The Indenture: The Contract Behind the Bond and Its Terms

An indenture is the document that carries a bond's terms: what is owed, on which dates, at what rate, how the amount borrowed comes back, what the borrower must and must not do meanwhile, and what counts as breaking any of it. A term that is not written into it binds nobody. Holders act through a trustee rather than one at a time.

The written term and the trustee carry most of the weight, and both are worth slowing down on. A bond is a promise made once and then handed around. The person holding it in year nine may never have met the borrower, may have bought it from somebody who bought it from somebody else, and has no way of asking what anyone meant. So the promise has to be parked somewhere that survives all that handing around, in a form that reads the same to the ten thousandth holder as to the first. The document is that parking place. And because a promise nobody can act on is not much of a promise, the same document has to settle in advance who is entitled to do something when it is broken.

A bond document is not written for the day it is signed, and that surprises most people on a first reading. The document is written for the day something goes wrong. Writing for the bad day is why the language is so careful about definitions nobody expects to need, why it spends whole sections on procedure, and why the interesting clauses are the ones describing failure. A contract drafted for the good case would run to a single sheet.

So what exactly is an indenture?

An indenture is the obligation, written down, once, in a form that binds whoever holds the claim later as completely as it bound the first buyer. Nothing more mystical than that. In Indian practice the same object also travels under the word debentureIn Indian usage, a bond issued by a company, very often under a deed held for holders by a trustee. It points at the same kind of object as the word bond; which word turns up depends on the market and the paperwork. and, with it, a trust deed, and the plumbing differs a little between the two, but the object being described is the same one: a written obligation that can change hands.

The property that matters most is that the promise and the document are the same object, so reading the promise means reading the document and nothing else. Not a summary somebody has sent. Not the sentence in the press note. Not what was explained on a call. A summary, a press note and a call may all describe the document accurately, and often do, but none of them is the document, and where one of them differs from the document, the document is what stands.

The same point turns up somewhere less exotic. A small workshop that repairs sewing machines keeps its terms taped to the wall beside the counter: one week's turnaround, no refund once the machine is opened, nothing left beyond thirty days. The person at the desk will happily say that they usually manage it in three days and usually take things back. Both statements are probably true. Neither is on the wall. On the day there is a disagreement, what is on the wall is what gets read out, and the word usually turns out to have been carrying no weight at all.

A bond works the same way, with one addition that makes the written form not merely useful but structurally necessary: the claim moves. The workshop's customer is standing right there. A bond's holder in year nine is a stranger who bought the claim from somebody else years after it was issued, and who is entitled to exactly what the document says, no more and no less, whatever anybody told the first buyer.

One more consequence of that, which sounds pedantic and is not. Because the document is a single object rather than a set of understandings, it can be complete on some points and completely silent on others, and its silence is not a soft no. It is nothing at all. There is no clause working in the background, no reasonable expectation being enforced quietly, nobody topping the document up with what would have been fair. Reading a silence as a soft no is the single most expensive mistake made about bonds, and it comes up twice more below.

What kinds of term does the document carry?

A bond document can run to a hundred printed sides, and readers meeting one for the first time usually try to cope by memorising the names of clauses they have heard of. Memorising names fails quietly. The names differ between issues, between markets and between drafters, and an unfamiliar clause then leaves the reader with nowhere to put it. The way through is to stop learning names and start learning kinds. There are five of them, and every clause sits in one.

A tenancy agreement for a two room flat has exactly the same five kinds, and most readers have seen one. The rent and the day of the month it falls due. The deposit, when it comes back and how much of it. Where that deposit stands if the building changes hands or the landlord has other creditors. No subletting, no knocking down walls, keep the place insured. And finally, what counts as breaking the agreement and what the landlord may do about it. Payment, repayment, ranking, conduct, failure. A bond document is longer, better drafted and far more careful about definitions, and it is doing those same five jobs.

An unfamiliar name leaves a reader stuck while a familiar kind always has room for it, so learning the five kinds beats learning a list of named clauses. Run through them once properly.

Payment terms

Payment terms settle what is owed while the bond is running. Three things have to be pinned: the rate, the base that rate is struck on, and the dates. Miss any one and the clause does not compute. A rate on its own is not a payment term; it is a number waiting for a base. Take the invented ten year 8.50 per cent bond that runs through this sequence. Its payment terms read as a rate of 8.50 per cent a year, struck on Rs 1,000/- of face amount, falling due on ten dates one year apart, with annual compounding throughout, which the document has to say because the same stated rate reckoned twice a year would work out to a different amount of money.

Repayment terms

Repayment terms settle what comes back at the end: the amount, the date, and whether anybody is entitled to bring that date forward. The third of those deserves attention. A right to repay early is a term like any other, and its absence is a fact about the document rather than a gap in it. Look at the ten year 8.50 per cent bond. Its repayment row puts Rs 1,000/- on the tenth date, which is why that date carries Rs 1,085.00/- once the last coupon amount lands alongside it, and the record gives it no early repayment right whatever.

Ranking terms

A ranking term settles rankingThe order in which competing claims on one borrower get paid when there is not enough money to pay all of them. It has its own treatment elsewhere and is not worked through here.: where this claim sits relative to everything else the borrower has borrowed, and whether any particular asset has been set aside behind it by way of a chargeA claim registered over a named asset, so that the asset stands behind one particular lender rather than behind everybody equally.. A claim with nothing set aside behind it is unsecuredBacked by the borrower's word and nothing put aside. A holder then stands alongside the general run of lenders rather than in front of them., which is not a defect and is simply a fact to establish rather than assume. Where a shortfall actually lands, and in what order, is covered separately.

Conduct terms

Conduct terms are the covenants: things the borrower undertakes to do, and things it undertakes not to do, for as long as the bond is outstanding. Keep insurance in place. Do not set aside assets behind somebody else. Stay inside a stated boundary on how much is borrowed altogether. A covenant nobody can measure cannot be broken and therefore does not do anything, so each one has to be testable against something stated.

Failure terms

Failure terms settle two questions that sound like one. First, what counts as the document having been broken, which is a definitional matter and is written out carefully precisely because it will be argued about. Second, what may then be set in motion. The follow-on steps are covered separately, under what happens once a bond term is broken.

The same five kinds, two different documents

A second instrument set beside the first shows how little the kinds change and how much the rows do. Palash Cements Limited borrows for five years at 9.10 per cent a year on the same Rs 1,000/- of face amount, so its payment row reads Rs 91.00/- on each of five dates and its repayment row puts Rs 1,000/- on the fifth of them, which therefore carries Rs 1,091.00/-. Counted across its whole run, that document contracts for Rs 1,455.00/- against the ten year bond's Rs 1,850.00/-.

Now the part worth carrying away. Both documents fill the payment kind and the repayment kind, and this record leaves the other three kinds empty on both instruments. Two filled of five, three empty, twice over. Three empty kinds are not a shortcoming of the record; they are the ordinary condition of a reader handed the headline terms and nothing else, and they are why the difference between a row that has been read and a row that has not is worth being insistent about.

There is one more figure floating around these two instruments that is worth pinning down here, because it is the kind of number readers assume must be written somewhere. Palash Cements borrows at 9.10 per cent a year. The sovereign borrows for the same five years at 6.90 per cent a year on this record's own spot curve, annual compounding. Set one against the other and the difference is 2.20 percentage points, which is 220 basis points. The 220 is arithmetic and not a term: a reader with both numbers in front of them works it out, and no clause anywhere creates it. The document holds 9.10 per cent a year. The market holds the rest. Read the two documents side by side and only sixty basis points divides Palash Cements Limited from the ten year 8.50 per cent bond, and that gap too is arithmetic performed by a reader rather than a term anybody agreed to. The purpose of a spread, and what it pays for, is covered separately.

Why is there a trustee, and what is the trustee for?

The reason is structural rather than legal, and once the structure is visible the legal machinery stops looking arbitrary. A bond is one borrowing, split into small pieces and sold to a great many people who do not know each other. Some bought at issue. Some bought last month. Some hold a hundred rupees of it and some hold a crore. Holders have different tax positions, different time horizons and, on any given day, flatly different opinions about what should be done.

Now let the borrower break something in the document. If every holder were free to act on their own, the outcome would not be decided by what is best for holders. The outcome would be decided by whoever moved first. One holder settles for a quick partial payment; the terms of that settlement now shape what everybody else can get. Another holder pushes for the harshest available step; the borrower's position deteriorates and the remaining holders inherit the wreckage. Where thousands of people hold one claim, individual action does not give each of them their own outcome, it gives all of them the first mover's outcome.

The same problem appears in a housing society. Three hundred flats, one leaking common roof, one builder who has gone quiet. If each owner chases the builder separately, the owner who settles first takes whatever is on the table, and the settlement terms then sit there as a precedent for the other two hundred and ninety nine, none of whom agreed to them. The society acts as one body for exactly that reason: not because the owners are barred from acting, but because scattered action produces a worse result for all of them than co-ordinated action does.

So the document appoints somebody to act for holders as a class. The trustee is named in the document itself, and the role matters more than the name.

The arrangement is a trade rather than a rule, and stating it as a trade is the honest way round: holders gain a voice that can actually set something in motion, and each one gives up the right to move on their own. Readers who meet this arrangement for the first time sometimes read the second half as a restriction imposed on them by somebody else. It is not. Giving up the right to move alone is the price of an instrument thousands of people can hold at once. A bond without the arrangement would either be unenforceable in practice or would have to be sold to one lender only, at which point it stops being a bond and becomes a loan.

Two things this party is not, both of which readers assume it into being. The trustee is not a guarantor: it does not stand behind the money and it does not pay when the borrower cannot. And it is not managing a holder's investment, deciding what is best and adjusting as it goes. The trustee acts on the document's terms. Those terms therefore have to be written in a way somebody can test. A clause saying the borrower will behave prudently gives an appointed party nothing to act on. A clause with a stated boundary gives it a fact to check.

Who is permitted to take that appointment, what duties come with it, and what it must do and when, are all settled by the Securities and Exchange Board of India (SEBI) at sebi.gov.in.

What happens to a term that is not written into the document?

Nothing happens to it: an unwritten term is not a term at all, an obligation is exactly what the document says, and everything else is conversation. The point is worth being unpleasant about. Every way of getting it wrong feels reasonable at the time.

Not a term that is usual. Not a term the borrower mentioned in a call, however sincerely. Not a term another bond from the same borrower carries, however recently it was issued and however similar it looks. Not a term assumed to be present because every bond met before had one. And not a term that would obviously have been fair. Fairness is not a clause, and nobody is quietly enforcing it in the background.

The mistake has a shape, and the shape is that people run one test when there are two. The test that feels natural is: is this the sort of clause bonds carry? The usual-clause question has an answer, and the answer settles precisely nothing about the document in hand. The test that decides is: is this clause written in this document? Only the second one is live. The first is a dead end that can be walked a long way down before the absence of an exit becomes obvious.

Sit with the last line of that drawing for a moment, because it is the part people skip. When a document says nothing about something, that silence is not a soft no and it is certainly not an implied yes. It is nothing at all. There is no clause operating quietly in the background, no reasonable expectation filling the space, nobody topping the document up with what most people would have agreed to. The document is complete on what it covers and blank on what it does not, and blank means blank.

The everyday version is a stall on a busy street selling phone chargers. The shop three doors down takes returns within a week and says so on a card by the till. The charger stall has no card. A customer may reasonably guess it behaves the same way, and it may well do so most of the time, and none of that is a term binding it on the day it decides otherwise. The guess was about the street. The obligation was about the stall.

Can a term be changed after the bond has been issued?

Yes, and the way it is done shows what a term actually is. A document sets out, inside itself, the route by which it may be altered. The route runs through the holders. Somebody proposes a change, the procedure written into the document gets read, holders are convened on whatever notice the document requires, and the change carries only if the agreement the document specifies is reached. If it is not reached, the term stands exactly as it was.

A term the borrower could change on its own would not be a term, and that is the entire reason the route is deliberately awkward. Think about what the alternative would look like. If the borrower could rewrite the payment date by announcing a new one, the payment date was never an obligation, it was a preference. The difficulty of changing it is not friction to be engineered away; it is the property that makes the written thing worth anything at all.

Two pieces of vocabulary turn up in this corner. A meeting needs a quorumThe minimum turnout a meeting needs before whatever it decides counts as a decision at all. Set below that, a meeting can be held and still settle nothing. before whatever it decides counts, and holders may sometimes agree to a waiverHolders agreeing, on the terms the document lays down, not to act on something that has already gone against them. It sets aside a consequence without rewriting the clause. rather than a change. A waiver sets aside a consequence without touching the wording. Both are decisions taken by holders rather than by the borrower, and that is the point. Notice, the convening of holders and the variation of a term once a bond has been issued are all set by SEBI at sebi.gov.in.

What does the card for this bond actually say?

The card below sets out the ten year 8.50 per cent bond, one row per thing that would need to be settled, and the honest version of what this teaching record holds. The card works as a template to carry to a real document rather than as a specimen of one.

Eight rows carry everything this teaching record holds. Four of those eight say something and four of them say nothing, and the four saying nothing are the interesting ones. Then a rule, and below it six more rows left empty, each carrying the body that settles it instead of an answer. Fourteen rows altogether, four of them filled.

Work down it slowly. The second half of that card is doing something unusual. The four rows carrying figures are the ones this sequence has already settled: the amount the rate is struck on, the rate itself, the ten dates with Rs 85.00/- falling due on each, and the tenth date carrying Rs 1,085.00/- once the last payment and the repayment land together. Nothing new there.

The four rows carrying an absence are where the teaching is. No borrower is named, and the row says so rather than guessing. There is no right to repay early, and that reads NONE rather than being left blank, because none and not known are different answers to different questions. No asset has been set aside, and that reads NOT STATED, which is honestly weaker than NONE. And there is no grade from any firm that publishes a rating scale: a grade is defined by whoever publishes the scale, in that firm's own methodology.

A card with four of its eight recorded rows carrying an absence is worth more than a card with all eight filled, because blanks can be audited and an invention cannot. A complete looking card gives nothing to check. The four blanks supply four specific questions to take to the actual document.

The ten year 8.50 per cent bond has no borrower standing behind it, and its borrower row reads not named for that reason. The zero coupon bond met elsewhere in this sequence has no borrower either. Its payment rows carry no dates at all and its repayment row carries exactly one. The amortising variant has none either. Its repayment row pays the amount borrowed down in slices instead of returning it in a single sum on one date. Palash Cements Limited and the five year government spot rate quoted alongside it were both written for the arithmetic and for nothing else. No grade attaches to any of them, and none of these figures is what anything is priced at or pays.

What can the document not do, however well it is written?

The document cannot make the borrower able to pay. The honest limit of the whole apparatus sits right there, and over-reading a well drafted document is a more expensive habit than under-reading a bad one.

Look at what a conduct term actually is. A conduct term requires the borrower to stay inside some stated boundary, and it says what may be set in motion if the borrower does not. Both halves are about consequences. Neither half is about capacity. A covenant is a trigger, not money, so it changes what happens next without changing by a single rupee the amount the borrower can pay.

Here is the shape of it drawn twice, once on either side of a boundary being crossed. Notice what moves and what does not.

A covenant genuinely buys time and information, and both are worth having. It buys information because a boundary that has to be tested against something stated forces the borrower to report against it, so somebody finds out early. It buys time because the moment a boundary is crossed, holders have an entitlement to act while there is still something to act on rather than after everything has gone. Neither of those is the same thing as being paid. The steps that follow a trigger, what holders may then do, and what portion of the amount owed a lender ends up seeing by way of recoveryThe portion of what was owed that a lender ends up actually collecting after a borrower has failed to pay. It is covered separately. after a borrower reaches insolvencyThe state a borrower reaches when claims falling due cannot be met from what is available. What follows from it is covered separately. are each covered separately.

The reader who has misunderstood this reads a tightly drafted document and concludes the money is safe. The reader has swapped a warning system for a guarantee. A smoke alarm in a kitchen is a good thing to have, it improves the odds considerably, and it has never once put out a fire.

How does anybody actually use this?

Four different readers pick up the same document and go to different rows first, and watching where each one goes is a fast way to understand what the rows are for.

A lender already exposed to the same borrower goes to the ranking kind before anything else, because that is where its own position gets decided by somebody else's paperwork. If this new bond has an asset set aside behind it and the existing loan does not, the existing lender's position has moved without the existing lender doing anything. Conduct terms restricting what a borrower may set aside behind future lenders exist for exactly that reason: they are one lender protecting itself against the next document.

A credit analyst builds the card before building an opinion. Not the rate first: the kinds. Which of the five are filled, which are empty, and what the empty ones would have to say before the opinion changes. An analyst who leads with the rate has an opinion about a number rather than about an obligation, and the two come apart quickly.

A household buying a bond through a distributor has less time and fewer printed sides in front of them, so the useful habit is narrow: read the repayment kind and the failure kind before reading the rate. The rate is the part everybody already reads and the part everybody quotes. When money comes back, and what counts as it not coming back, are the two things nobody mentions and the two things that decide the outcome. The routine is a narrow one, and it is worth carrying away.

And an existing holder wanting to sell reads the document because the next buyer will. The buyer is paying for the set of rows, exactly as written, that will pass along with the claim, not for the borrower's reputation and not for the rate printed on the front sheet.

Who decides what has to be in the document, and who may enforce it?

Six separate questions sit behind that heading, all six drawn as empty rows on the card and listed again in the block beneath it. The contents of the trust deed behind an issue. Who is permitted to take the trustee appointment and what duties come with it. The disclosure required when a bond is offered. The security that may be created over a borrower's assets, and how it must be registered. How holders must be convened and how a term may be varied once the bond is out. And how a grade, where one exists, must be disclosed and kept under review.

For a company's borrowing all six belong to SEBI at sebi.gov.in. Where the borrower is the sovereign, the equivalent ground is held by the Reserve Bank of India at rbi.org.in. Every one of the six gets revised, on a timetable that belongs to the body setting it and to nobody else.

The principle is the same one as the stall on the street. Talk about a bond is no substitute for the bond's own document, and a summary of a requirement is no substitute for the body that sets it.

The shape set out here is the durable part, and it is genuinely durable. A document exists, everything binding lives inside it, its clauses sort into five kinds, holders act as one class, a term not written down binds nobody, changing a term runs through the holders, and no clause creates money: none of that changes when a circular is reissued. The content of those six rows changes. With the shape learned here and the content read at its source, a revision dates the routing rather than the understanding.

One last thing about enforcement is worth separating out. Two different things are going on and they get run together. Making sure the document contains what it must, and that what has to be disclosed was disclosed, sits with the bodies named above. Acting on the terms once they exist sits with holders, exercised through the party appointed for them, on the route the document itself lays down. The first is supervision of the paperwork. The second is the paperwork being used. Neither substitutes for the other, and what actually follows once a term has been broken is covered separately.

The ten year 8.50 per cent bond, the zero coupon bond, the amortising variant, Palash Cements Limited and the five year government spot rate standing beside it are invented.
Educational material. Not advice on any investment, tax, budget or market position.