How to Write a Fixed Income Note a Reader Can Check

Eight parts, in one order. The claim comes first and complete, with its conditions in the same paragraph. Then the figures, each carrying its base, its period and its unit. Then every assumption named where it is used, every missing figure printed as a labelled blank, one sentence naming what would overturn the claim, the limits, and a full stop.

A note is read once, in one direction, by somebody who very often stops after the opening paragraph. One reading in one direction settles the order of everything in it. The single pass also stops the order of a note from matching the order in which the work was done. The writer arrives at the claim last, after the arithmetic and the arguing, and the reader has to be handed it first.

The eight parts, in order

1. The claim, complete, in one sentenceIt has to survive being lifted out and read alone. Not the background, not the method, not what the borrower makes for a living.
   Cover everything below it and read it. If it needs the heading to make sense, it is not the claim yet.
2. The conditions, in the same paragraphTwo or three things that would have to be true for the claim to hold, written as part of the same thought rather than gathered up at the foot.
   Read the opening paragraph on its own. Does it still say what has to be true?
3. The figures, each carrying four labelsThe base it is struck on, the length of time it covers, the unit it is in, and the word saying which object it is.
   A reader handed the note and a calculator must land on the same number without asking a single question.
4. The assumptions, named where they are usedIn the sentence that leans on them, every time, rather than collected once in a schedule at the back.
   Find each figure that came out of an assumption. Is the assumption within a line of it?
5. The missing figures, printed as blanksIn the place the figure would have gone, with the reason for its absence sitting inside the blank.
   The questions the note cannot answer are worth counting, and each one has to be visible in the note itself.
6. One sentence saying what would overturn the claimA specific observation, named before anybody asks for it, that would make the writer write the note differently.
   Could a reader watch for it? A sentence nobody can watch is not this sentence.
7. The limits, and only then the disclosuresWhat the work rests on and how thin that is, followed by the block whose wording belongs to somebody else.
   Is the disclosure block present, labelled, and empty rather than filled from memory?
8. The stopThe note ends. It says on its face that it is ending, and it names nothing anybody should go and do.
   Is the last sentence an ending, or is it an instruction wearing an ending?

What goes first in a note, and why nothing else may?

A neighbour messages the residents about the overhead tank. If the message opens with the year the building went up and the name of the contractor, it gets scrolled past. If it opens with the words the tank is leaking into the stairwell, somebody puts their shoes on. Neither message is untrue and both contain the same facts. Only one of them works on a person who is going to read a single line.

A fixed income note opens with the thing it was written to say, complete, in one sentence that survives being lifted out and quoted on its own. Not the background. Not the method. Not a paragraph on the borrower's line of business, however interesting that is and however much of the writer's week it consumed. The reader who stops after that sentence must not be left holding something misleading, and the reader who never returns must still have the point.

There is a mechanical test for it, and it takes four seconds. The test is to cover everything below the opening sentence and read what is left. If it needs the heading to make sense, it is not the claim yet. If it needs the next paragraph, it is not the claim yet. If a colleague could paste it into a message with no surrounding text and it would still mean the same thing, the claim is there.

The practice of putting the claim ahead of everything that supports it is usually named for Barbara Minto, and the name travels with it because she is the one who set it out as a discipline rather than a preference. The rest is ordinary professional care, and it belongs to nobody.

Difficulty with this rule has nothing to do with subtlety. The writer discovers the claim last. The writer reads the terms, works the arithmetic, notices what is absent, argues with somebody, and only then knows what to think. Writing in that order is natural, and it produces a note whose point arrives on the third printed side, by which time it is arriving to nobody.

Where do the conditions of the claim go?

A caterer quotes a figure for the wedding. The figure holds at three hundred guests, on a weekday, with the pandal already up. If the figure sits in the first line and the three conditions sit on the last sheet of the quotation, the family will remember the figure and forget all three, and so will everyone they repeat it to. The caterer has not lied. The caterer has simply put the number where it travels and the caveatA warning fixed to a statement, marking where the statement stops being reliable. where it does not.

A claim in fixed income research is almost never unconditional. Conditionality is not a weakness in the subject and it is not a hedge. An honest reading of a price looks exactly like that. The conditions belong inside the same paragraph as the claim, written as what would have to be true. Claim and conditions are then one thought, and they travel as one thought.

The reason is not thoroughness and it is not politeness to the reader. The opening paragraph is the part of a note that gets quoted. The quoted words travel into a message, into a summary, into somebody else's slide. Everything below it stays where it was put. A note that states a claim on its first printed side and its conditions on the fourth has written a strong first paragraph and a weak document, and the difference will only show up later, in somebody else's mouth.

Two or three conditions is the usual shape, and each one is a qualifierA word or phrase that narrows a statement, so it claims less than the bare sentence would have claimed. doing real work rather than a piece of protective wording. The test is whether removing it changes what the claim says. If it does not, it is padding and it is costing the reader's attention. If it does, it has to sit exactly where the claim sits.

What must every figure in the note carry with it?

Four labels and one test, applied to every number without exception, including the ones inside tables and the ones inside captions. The four labels are cheap and mechanical. The test is the one that catches what the labels miss.

The fourth label is the first casualty of a tired afternoon, and it deserves a paragraph of its own. A rate on this platform always says SPOT or says FORWARD, and the reason is that a one year rate one year FORWARD and a three year SPOT rate can sit a handful of basis points apart while being entirely different objects. A duration always says MACAULAY or says MODIFIED. The two differ by a factor, and a note that mixes them is out by exactly that factor. A coverage reading always says gross or says net. None of these words is decoration; each one is the difference between a figure a reader can place and a figure a reader has to guess at.

Now the test. Being rerunnable is a separate requirement from being right, and it is the one almost never checked. Every figure in the note must be reachable from the other figures in the note. Somebody who has never met the borrower, holding the note and a calculator, must be able to walk from the inputs to the answer without putting a single question to the writer. If they cannot, the note has a defect in it whether or not the answer is correct, and the defect will be found by somebody a good deal less friendly than the reader the writer imagined.

The failure has another side to it. A reader recomputes, lands somewhere else, and does not write to the analyst. The reader assumes the misunderstanding is theirs, loses a little confidence in their own arithmetic, and quietly stops recomputing anything from that desk. The writer never hears about it. The silence is the cost, and it compounds.

There is a smaller version of the same fault, and it hides inside a rounding. The specimen below prints 3.67 per cent as its implied yearly rate of default. Run the arithmetic backwards on the unrounded 3.666667 per cent and it returns 2.20 percentage points exactly. Run it backwards on the printed 3.67 per cent, against the same 60 per cent severityThe part of what is owed that stays lost after a failure to pay. Whatever recovery somebody has assumed, this is the rest of it., and it returns 2.2020 percentage points. The gap is 0.20 of a basis point from where it started. Nothing is wrong with either the input or the output. The mistake is treating a display figure as though it were an input.

Where does an assumption get named, and why not in a list at the end?

A tailor tells a customer the shirt will be ready on Friday. The tailor has not said that Friday depends on the cloth arriving from the wholesaler on Tuesday. The customer does not hear a conditional. The customer hears Friday, plans around Friday, and when Friday does not happen feels misled by somebody who was, in their own mind, being perfectly clear.

An assumption is named in the sentence that leans on it, every time that sentence appears, and not once in a schedule the reader will not reach. The reason is not thoroughness, and it is not a matter of covering yourself. A figure produced by an assumption starts reading as a measurement the moment the assumption is more than a line away from it. The gap does the damage, not the omission.

The person who actually quotes a figure is moving quickly, working off the paragraph in front of them and reading no further. The material has been reliable before. Such a reader is precisely the one least likely to turn to the back and check what the figure rested on. Pushing the assumption into an appendixA section pushed to the back of a document to hold supporting material. Readers reach it last, when they reach it at all. arranges for the one person who most needs the qualifier to be the one person who will not see it.

Written properly, the assumption and the figure share a single sentence. Recovery is ASSUMED here at 40 per cent, severity therefore stands at 60 per cent, and 220 basis points of spread carried across that severity point to 3.67 per cent a year, implied. Every clause there is carrying weight. The word assumed is doing as much work as the arithmetic. Move it to the back and the sentence still contains 3.67 per cent, but the 3.67 per cent has quietly changed what it appears to be.

Notice also the word implied, welded to the rate wherever the rate goes. An implied rate reports where a price sits, given an assumption somebody chose. No forecast is being made. Nothing is being counted either, least of all how often issuers have actually failed. Strip the word away and a reader takes the number for a measured frequency. A measured frequency is a far stronger claim than any note on this platform is entitled to make.

How is a missing figure written into a note?

A rent agreement leaves the electricity meter reading blank, with the words to be read on the day of handover written across the space. Nobody is confused by that blank. Everybody knows exactly which fact is outstanding and when it will land. The same agreement with the meter line simply left out passes unnoticed, and a reader who did notice it would assume somebody had checked the point and found nothing worth saying.

A missing figure goes into the note in the place the figure would have occupied, printed as a blank, with the reason for its absence sitting inside the blank, in the same size type as everything around it. Three practices sit around that blank. All three are real, all three get reached for at four in the afternoon, and none of them is the note.

A labelled blank is not an omission. An omission is invisible, and invisible is worse than absent. The reader fills the gap with the assumption that the writer considered the point and found it unremarkable. Nor is it a figure borrowed from something similar. Borrowed figures cannot be told apart from measurements once the note is a fortnight old, and by then it is being quoted. And it is not a phrase such as broadly in line. Such a phrase is a filled blank wearing words: it occupies the space, it survives review, and it carries no information at all.

The specimen note further down has three places where a reader would reasonably expect a figure and where nothing on this platform can supply one. All three are printed as labelled blanks. The first is what the instrument would fetch if somebody sold it, and no trade is observed anywhere in this material. The second is how often issuers of this kind have failed, and no history of failures is held anywhere in it. The third is what any other issuer pays for this maturity, and a single issuer stands alone with nothing to be set beside it. Three blanks, three different reasons, and a reader who now knows precisely which questions the note cannot answer.

The last of the three is worth holding on to. A blank written to that standard is not an apology and it is not a gap in the work. The blank carries information of its own. A reader learns exactly where the edge of the evidence sits, and most notes never manage to say that at all.

What does the note say about what would change the writer's mind?

One sentence. Written before anybody asks for it, naming the observation that would make the writer revise the claim. The sentence is the cheapest in the note to write and the one most often missing. Its absence is the clearest signal available that the reader is holding advocacyWriting meant to bring a reader round to a position, rather than to put them in a position to test it. rather than research.

The sentence has to be specific enough that somebody could actually watch for it. Conditions could deteriorate is not the sentence. The phrase names nothing, it can be satisfied by anything, and a year later nobody can say whether it happened. The sentence needs an observation with a number attached and a base attached. A reader can then put it on a list and check it.

On the specimen note the sentence names a level on the coverage reading. Gross interest coverage stands at 3.4000 times for the year, struck on earnings before interest and tax of Rs 154.70 crore over interest expense of Rs 45.50 crore. The note declares a watch level of 3.0000 times, computed on that same base over that same length of time. Below it, the claim above no longer holds and has to be rewritten.

A level in times is hard for anybody to keep an eye on, so the watch level reads more easily in rupees. Interest expense of Rs 45.50 crore multiplied by 3.0000 times is Rs 136.50 crore, so the watch level is reached when earnings fall from Rs 154.70 crore to Rs 136.50 crore. The fall is Rs 18.20 crore. Set against the earnings themselves it is one rupee in every eight and a half. Nothing lucky happened there. Dividing 3.4000 by 0.4000 gives the same 8.5. The interest expense drops out of both halves of the ratio.

The watch level of 3.0000 times is declared in this guide for the specimen and is not observed anywhere and not required by anybody. The level was chosen to be round, visible and below the reading. Where the level sits matters less than that it sits somewhere a reader can point at, with its base and its length of time matching, exactly, those of the figure it is watching. A level computed on a different base is not a watch level; it is a second figure pretending to be one.

Where does the note stop, and what is deliberately not in it?

A plumber looks at the inlet line, tells the householder the joint is corroded, tells them what it will cost to replace, and tells them the estimate assumes the wall does not have to come down. Then the plumber stops. The plumber does not say which brand of tap to buy. A plumber who did would be thought slightly less of. The brand was not what was asked and not what the plumber examined.

The note stops once it has given the claim, the conditions, the figures, the assumptions, the blanks, the revision condition and the limits, and it says on its own face that it is stopping. Four things stay out. An instruction to do anything. A target for any price. An adjective standing in where an argument should have been. And any requirement set by an authority, written from memory.

The fourth is the quiet danger. Writing a requirement from memory does not feel like a risk while it is being done. A disclosure duty or a conflict rule feels like something a professional simply knows, in the way anybody knows their own address. A rule of that kind also arrives in a note as boilerplateStandard wording carried into every document unchanged, so that after a while nobody rereads it, including the person who put it there., copied forward from the last note, and that one was copied forward from the one before. The disclosures a published view must carry, who counts as conflicted, who is permitted to publish and which records must be kept are all set and revised by somebody else. Write one of them down from memory and the sentence is not merely dated when it changes; it is wrong, sitting inside a document somebody is relying on.

The move is the same every time, and it takes one line. Name the item. Name who keeps it. Leave the space empty. A labelled empty space with an address beside it serves a reader better than a confident line copied out of somebody's recollection. The empty space also cannot go stale. The confident line will.

Whether any given sentence belongs at all comes down to one question with three answers rather than two. Can a reader check it without asking the writer? If yes, it goes in as a fact with its source beside it. If no, it goes in as an assumption with the word attached. And if the sentence tells the reader to go and do something, it does not go in at all. Instructions belong to a different kind of document.

What does the finished note look like when every part is present?

Here it is, whole, on the invented issuer, with each part marked in the margin. The specimen is deliberately short. A reader who has just been handed eight rules needs to see all eight in one object small enough to hold in the eye at once, and a long example would teach length instead of structure.

Twenty nine lines, eight parts, and the shape of it is worth looking at on its own. Six of the eight parts are two or three lines each. The largest part is the one carrying no content at all, the disclosures block, and no accident of this specimen produced it. An empty block that is honestly labelled takes as much room as a full one and is worth having anyway.

No slider appears below, and the reason sits in the subject rather than in the effort. The eight parts vary only in ORDER, and order does not arrive in amounts. A claim shifted from the opening paragraph to the fourth printed side has not moved 40 per cent of the way anywhere: a reader either holds the claim or does not. A control that stripped parts out of a note one at a time would put an animation on a checklist, and an animated checklist is decoration wearing the clothes of a model. The specimen note at the foot of this guide does the work a redraw would have done, and it survives being read out loud. No slider does. In its place there is a ninth question.

Who actually reads a note this way, and what do they do with it?

An analyst at a lender opens an outside note on a borrower the bank already has money with. She is not reading it the way it was written. She reads the opening paragraph, and then she goes hunting for exactly two sentences. The first is the one naming what the figures assumed. The second is the one naming what would change the writer's mind. If she finds both, the note goes into the credit file as evidence and she watches for the thing it named. If the second is missing, the note goes into the file marked as somebody's view, and it stops carrying weight in any decision.

The sorting takes her under a minute, and it is done before she has read a single one of the tables. Everything the week was spent on sits below the line where the sorting happens.

The same reading habit belongs outside finance and is worth practising there, where the stakes are smaller and the shape is identical. A household getting three estimates for a roof repair should read each one for the same two things: what has this person assumed, and what would make them change the figure. The estimate that says the price holds if the beams are sound, and that a rotten beam would move it, is worth more than the one with the lower number and nothing else in it. One of them can be checked as the work proceeds. The other can only be believed or not.

There is one more use, and it is the private one. A revision sentence written honestly gives the writer something to check against later. Six months on, what was named can be looked at again to find out whether it was wrong. Nobody gets better at this any other way. A writer who never names a breaking point has arranged never to receive that information, and will keep writing the same note for years.

Palash Cements Limited is invented.
Educational material. Not advice on any investment, tax, budget or market position.