The Payment Date: When the Money Actually Moves

The payment date is the day money actually moves between the two sides of a swap. Each leg is worked out in full for the period that has just finished, and then a single transfer settles both of them, sized at whatever separates the two amounts. The figure both legs were multiplied by stays exactly where it is. Everything else in the agreement is a rate, a date, or a way of counting.

A swap agreement, read plainly, is a long set of instructions for arriving at one number on one day. The rates settle what each side owes. The counting method settles how much of a year the period was. The calendar settles when the answer falls due. The payment date is where all three of those arrive together and stop being instructions, and no other date in the document leaves anybody genuinely out of pocket. Every other date decides something. The payment date pays for it.

At the head of the agreement below sit two names. Chitrakoot Cements Limited, an invented manufacturer, has undertaken to hand across a rate that never shifts for the life of the document, 7.20 per cent a year, and to take back in exchange whatever the floating benchmark happens to read. Saranga Capital Limited, an invented lender, signed the mirror of that sentence: out goes the floating benchmark, back comes the fixed rate. Written underneath both signatures is a notionalThe figure both legs get multiplied by. The notional fixes how large each payment is and is not itself handed over by anybody. of Rs 1,000 crore, and neither side has parted with a paisa of it.

Behind any statement of what a swap is worth today sits a list of what the floating benchmark is expected to read on every future date in the schedule. Without that list the first period can still be worked out to the rupee, and every period after it has a real date on the calendar with an empty box sitting above that date. A single reading fixes one period and no more, and an empty box is the honest mark for the rest.

Here is the same idea with none of this vocabulary in it. An electricity account has three dates on it, and they rarely need to be thought about separately. Somebody reads the meter on one day. A clerk or a machine turns that reading into a bill on another. And on a third day the money leaves the bank account. Only the third one shows up on the bank statement. Asked when the electricity bill was paid, nobody names the meter reading date, even though the meter reading decided the amount. Naming the day the money left is exactly right. The day the number gets decided and the day the money moves are different days, and they are different on purpose.

Four kinds of date sit in one agreement. Which of them moves money?

A swap document is dense with dates and a reader who cannot separate them cannot read the arrangement at all. There are four kinds. Mixing any two of them produces a confident wrong answer rather than an obvious mistake, and a confident wrong answer is expensive.

The trade date is the day the two sides agree. Nothing is computed on it and nothing is paid on it. The terms stop being negotiable: the fixed rate goes into the document, the notional goes into the document, the counting method goes into the document, and the calendar of future dates goes into the document. From that moment the agreement is an instruction sheet that both sides have signed.

The effective date is the day the first period starts running. Paperwork between two counterpartiesThe other side of a privately agreed contract, as opposed to a crowd of strangers meeting on an exchange. takes a few days to complete, and the effective date usually sits a short distance after the trade date. Again nothing is paid. The clock starts, and the counting method will eventually count it.

A reset dateThe day a floating leg picks up the number it will be charged on for the period about to start. The reading itself is covered separately. is the day the floating leg takes its number for a coming period. There is one of these for every period, so they recur all the way down the schedule. Something important happens on a reset date, but it is an act of reading rather than an act of paying: a figure is observed, written down, and applied to the period ahead. The mechanics of taking that reading are covered separately, and the floating amount for the period worked below is already known.

And the payment date is the day money moves. On every other date in the document something is decided, and on a payment date something is transferred. That single line is worth more than any list, because it survives being applied to an unfamiliar document. The dates on which something is transferred are the only dates on which either side is out of pocket.

What has to be settled before a payment date can be paid at all?

Three things, and every one of them has to be nailed down before anybody can send anything. Take them in the order the document takes them.

First, the rate that applies on each leg for the period that has just ended. The fixed rate was typed into the agreement and it does not move, so on the fixed leg the rate has been known since the trade date. On the floating leg it was known from that period's reset date, when the benchmark was read.

Second, the day count fractionHow large a slice of a year an agreement treats one period as covering, arrived at by counting its days the way the document instructs. The counting itself is covered separately., which is how much of a year the period is treated as being. Every rate in the document is quoted for a year, and almost no period is a year long, so something has to translate between the two. The period worked below is one full period at a fraction of 1.0000, and that fraction is repeated beside every figure rather than stated once at the top. Arithmetic that is not shown is arithmetic nobody can check.

Third, the figure both legs are computed on. The notional has been sitting in the document since the trade date doing nothing except waiting to be multiplied.

Two of those four were settled on the day the agreement was signed, and only the floating reading had to wait. Waiting for that one reading is the entire reason a payment date sits at the end of a period rather than at its start. Nothing else in the document is waiting for anything. If the floating leg had also been fixed on day one, there would be no reason at all not to pay on the first morning, and the arrangement would be a loan schedule rather than a swap.

Why is the amount worked out at the end of the period rather than the start?

Because two of the three inputs are only complete once the period has finished. The floating leg follows a benchmark reading taken at the start, and that much is settled early. But the period then has to run its full length before anybody can say how many days it actually contained, and therefore what share of a year to apply to both rates. Until the last day of the period has happened, the fraction is a plan rather than a fact.

Working the amount out at the end of the period is what practitioners call paying in arrearsWorked out and settled at the finish of the stretch of time it relates to, rather than at the front of it.. The phrase sounds like jargon and is not: it is simply a description of which end of the period the money sits at. An electricity account works the same way, and so does a salary. Nobody is paid in January for work not yet done in January.

A party knows the rate it will be charged on before a period begins, and does not know the amount until that period has finished, and those are two entirely different kinds of certainty. Treating them as one is how a treasury forecast goes wrong without a single arithmetic error anywhere in it. The rate really was known, so somebody writes down a figure in October for a March payment date and feels reasonable about it. The number was never knowable. The figure depended on a length of time that had not yet elapsed.

Notice how mild the failure is and how completely it hides. There is no wrong formula to find. There is no reversed sign. There is a confident figure produced from an input that was not yet a fact. Everybody who looks at such a figure can see where each part came from, and that is what makes it the hardest kind to catch.

How do two computed amounts turn into one transfer?

Turning two computed amounts into one transfer is the step worth taking slowly. There are three movements in it, and skipping any of them leaves a picture that works on this agreement and fails on the next one.

Movement one: each leg is computed in full and separately, exactly as though it were going to be paid on its own. Nobody looks at the other leg while doing it. The fixed leg is the notional multiplied by 7.20 per cent a year over the period fraction. The floating leg is the same notional multiplied by whatever the benchmark read, over the same fraction. Each of those is a real, complete, grossAn amount standing on its own, before anything at all has been set against it. obligation on the day it is computed.

Movement two: the smaller is taken away from the larger. One number survives.

Movement three: a single transfer is made for that surviving number, running in the direction of whichever leg came out larger. Both obligations are then dischargedOwing nothing further for that period, with the obligation closed off and nothing left outstanding on either side. in full, even though only one amount travelled and it was smaller than either of the two amounts owed.

The worked instance, in four steps

Step one, the inputs that were already sitting in the document. A notional of Rs 10,00,00,00,000/-, or Rs 1,000 crore. A rate of 7.20 per cent a year on the fixed side, settled between Chitrakoot Cements and Saranga Capital when they traded. And one full opening period, counted as a day count fraction of 1.0000.

Step two, the one input that had to wait. Read for period one, the floating benchmark showed 6.00 per cent a year. The reading governs the opening period and nothing whatsoever beyond it.

Step three, the two legs, worked separately and in full. Put 7.20 per cent a year across Rs 10,00,00,00,000/- over a fraction of 1.0000, and the fixed side lands on Rs 72,00,00,000/-, or Rs 72.00 crore before anything at all is set against it. Send the same notional through 6.00 per cent over the identical fraction and the floating side settles at Rs 60,00,00,000/-, or Rs 60.00 crore standing alone. Two complete obligations, each worked as though it were about to be paid on its own.

Step four, the netting. Set Rs 60,00,00,000/- against Rs 72,00,00,000/- and Rs 12,00,00,000/- is left over, or Rs 12.00 crore of net difference. The fixed leg was the larger of the two, so one transfer of exactly that amount goes from Chitrakoot Cements to Saranga Capital. Both sides are then square for the period.

A shortcut sits in that table and it is worth naming. Spotting it and missing it lead to different mental models. Since both legs run on the same notional over the same fraction, the whole exercise collapses into applying the difference between the two rates to the notional once. The two rates are 1.20 percentage points apart, and Rs 1,000 crore taken through 1.20 percentage points is Rs 12,00,00,000/-, the same answer in one move. Flip the same relationship over, set the net of Rs 12.00 crore beside the notional it was drawn from, and 1.20 per cent comes straight back out, the rate gap read off the payment.

The shortcut gives the right answer and the long route gives the right understanding, so do not let the shortcut replace it. Two things break the shortcut immediately: legs computed on different notionals, and legs that fall due on different dates. Both are ordinary. A reader who only ever learned the one step version has learned a special case and does not know it is one.

One more thing happens in movement two that the drawing cannot show, and it is the part practitioners care about most. Both sides work out both legs independently and compare answers before anybody sends anything. That comparison is not ceremony. The comparison is the moment two organisations discover they disagree about how many days the period had, or which reading was taken, or what the fraction should have been. Discovering that disagreement while the number is still on a screen costs a phone call. Discovering it after a transfer has landed costs a reconciliation, an amendment, and somebody explaining to a finance committee why a settled figure came back.

Who pays whom, and was that settled when the document was signed?

Half of it was and half of it was not. The labels are so firm that all of it looks settled.

The agreement calls one side the fixed payer and the other the floating payer. The labels are permanent. Chitrakoot Cements is the fixed payer on the first payment date and on the last one, and nothing that happens in between changes that. The label describes which rate a side is charged on, not which way money goes.

The direction of the transfer, by contrast, is settled afresh every period, and it depends on nothing except which leg came out larger. The fixed payer sends money on a payment date only where the fixed rate produced the bigger amount for that period. Where the floating reading came out above the fixed rate instead, the fixed payer receives, and it receives without any amendment, any renegotiation or any change to what it is called.

A lot of confused reading starts here. Somebody looks at a document, sees the words fixed payer against a name, and writes into a note that this party pays every quarter. It does not. The party is charged on a fixed rate every quarter. The difference between those two statements is the difference between a cash outflow and a cash inflow.

One case is worth carrying away, and it looks like a fault without being one. Where the two legs come out equal, the difference is nil, and nothing at all moves on the payment date while both sides are nonetheless fully discharged for the period. A payment date on which nothing is paid is an ordinary payment date. The agreement worked exactly as written. Nobody missed anything.

Household version, and it is the same arithmetic. Two cousins share the cost of a car they both use. One has agreed to put in a flat Rs 6,000/- a month. The other puts in whatever the fuel actually came to. On the first of the month they compare the two figures and one of them hands over the gap. In a month of heavy driving the gap runs the other way and money comes back to the flat contributor, so nobody would describe that cousin as the one who pays. The label says what each of them is charged on. The month says which way the cash goes.

The relationship between a floating reading and the size and direction of the net difference is set out separately, where moving the reading redraws the line and flips the arrow. The calendar is what turns that relationship into a date, an amount and a direction.

What if the two legs do not fall due on the same day?

Most treatments skip this, and skipping it is what leaves readers with a tidy picture that quietly breaks. Nothing in the general form of a swap requires both legs to pay on the same calendar. An agreement can perfectly well have one leg falling due less often than the other, and plenty of them do.

Calendars that do not coincide do not weaken the netting step. Netting simply stops being available. Netting exists only where two amounts come due on one and the same date, so a date carrying just one leg sees that leg go across in full with nothing at all standing opposite it. No difference is taken. The gross amount goes.

Read back over everything above with that in mind and something uncomfortable becomes visible. The single tidy transfer, the small number instead of the two big ones, the instruction with three fields on it: all of that is what happens when the two calendars coincide. Coinciding calendars are the common case and the case worth learning first. The tidy transfer is not a property of the instrument. The tidy transfer is a consequence of a calendar.

The practical version of this is unglamorous and expensive. A treasury team that has learned the tidy picture funds an account for a difference. A date arrives on which only one leg is due, the full gross amount is called, and the account is short by the size of a leg. Nothing was mispriced. Nobody read a rate wrong. Two calendars simply did not line up, and the mental model had no room for that possibility.

What actually goes wrong on a payment date?

Somebody who runs settlements gives a very different list from the one somebody who studies rates would give, and a very different one from what most readers expect. The failures on a payment date are clerical, and the money lost that way is lost to process rather than to markets.

Look at the third column. Not one of these four has anything to do with what the benchmark did. Rates decided the size of the number long before the payment date arrived, and by the morning of the payment date the number is simply a figure on a document that has to reach the right account, on the right day, facing the right way.

The one that catches the most people is the second, so it deserves a sentence of its own. Counting is a convention rather than a fact of nature, and the convention lives in a clause somebody has to read. Two teams working from the same agreement can genuinely count the same stretch of time differently. A small shortfall arrives, and because it is small it gets parked. Parked shortfalls accumulate, and the reconciliation that eventually finds them is a great deal more expensive than the phone call that would have caught the first one.

Why can no amount be put against a later payment date?

Because a later amount rests on a benchmark reading that has not been taken, and the shape of that gap is worth teaching.

Every payment date after the first settles on a benchmark reading that has not been taken. A reading cannot be constructed, borrowed from a neighbouring period, or interpolated between two convenient points and still be a reading. An amount put against the fourth payment date would have been fabricated, and nobody reading it would have any way on earth of telling. The danger is exactly that. A fabricated figure does not look fabricated. The fabrication looks like every other figure beside it, sits in the same font, lines up in the same column, and gets copied into somebody's note with the same confidence as the one that was actually computed.

The later dates themselves are entirely real. The dates were fixed by the agreement on the trade date and nobody is in any doubt about them. Only the amounts are missing, and a date that has not been decided is a different thing from an amount that cannot yet be known.

Filling those boxes takes something specific and short: a benchmark reading for each later period, taken on that period's own reset date, published by whoever administers the benchmarkA published rate that a floating leg follows. Who administers one, and which may be referenced, is set by an authority and is not written out here., and carrying a date. Nothing less than that will do, and no reading can be produced from memory.

Who actually uses this, and what do they do with it?

Reading a payment date correctly is not a classroom skill. Four different people reach for it in ordinary work, and each of them is answering a different question with the same reading.

A treasury officer at Chitrakoot Cements is deciding how much has to sit in which account, and on which mornings, across the years the agreement still has to run. For this arrangement, on the coinciding calendars taught above, the answer is one modest amount per period and its direction is not known in advance. One modest amount of unknown direction is a very different funding plan from one full leg per period. Getting it wrong does not produce an analytical embarrassment that surfaces in a review. Getting it wrong produces an account that is short on a morning when a payment has to go out.

A credit analyst reading disclosures wants to know what this arrangement actually does to the company. Finding Rs 1,000 crore attached to it and treating that as an obligation produces a company buried in commitments it does not have. Finding one net difference per period, of unknown direction, produces something much closer to the truth. The test that separates the two readings is the one drawn above: look at the instruction that moves money and see whether the headline figure has a field on it.

A settlements clerk is the person for whom the payment date is the entire job. Their questions are the four in the table above and nothing else: is the date right, is the amount right, is the account right, is the direction right. The clerk is also the last line of defence against every one of those four failures, and usually the only person who computes both legs independently on the morning rather than accepting the figure that arrived.

And an auditor or reviewer comes at it backwards. The auditor has a bank statement showing Rs 12,00,00,000/- and a forecast showing Rs 72,00,00,000/-, and wants to know whether the company misforecast or misexecuted. Misforecasting and misexecuting are two entirely different findings with two entirely different remedies, and separating them takes exactly the reasoning set out above: the gross legs were both real, the netting was correct, and the forecast used a gross figure where a net one was going to settle.

The household version is smaller and identical in shape. Two households share a monthly tuition arrangement for a shared tutor: one pays a flat share, the other pays whatever the hours came to, and they settle the gap on the first of the month. The one who budgets their flat share as an outgoing every month is exactly the finance team in the failure block. Most months the money they actually part with is a fraction of that, and in a heavy month it comes back to them instead. Their bank balance is right and their plan has been quietly wrong all year.

What does knowing the payment date not establish?

Reading the calendar of an arrangement is a genuinely useful skill, and it is a reading skill and nothing more. The skill allows somebody to take an unfamiliar document, separate its four kinds of date, work both legs for a period, net them, and say which way the money goes. Saying which way the money goes is worth being able to do. Saying whether the arrangement suits anybody is a different question altogether.

Whether an arrangement of this kind fits a particular party turns on facts the agreement itself does not carry. Sensible, cheap, efficient or worth doing are not properties of the instrument. Each is a property of one party's own position.

Four things would have to be on the table before that question could get an honest answer. First, what the party already owes, and on what basis. Over what stretch of time, and whether the money arriving moves with the same benchmark as the money leaving. How it would cope if a payment date came round and the other side simply was not there. And what several consecutive periods running the wrong way would actually do to it. Not one of those four is visible in the agreement, and an arrangement recommended without them is recommended to a stranger whose circumstances are entirely invisible.

The deeper obstacle is arithmetic rather than policy. One agreement, one benchmark reading and one net difference make a single observation, not a realised result, a run of history, a probability or a spread of possibilities. Nothing can be ranked off a single observation, and phrasing a ranking carefully would not make it any more honest.

The arrangement puts a side out of pocket on a payment date whenever the period runs against that side. Being out of pocket is not a mistake somebody made, and not a sign that anything has gone wrong.

What is settled elsewhere

Working one date properly means leaving the neighbouring dates and mechanisms to their own treatments, where each gets the room a compressed preview could not give it. The neighbouring subjects, and the place each of them is handled, are these.

Chitrakoot Cements Limited and Saranga Capital Limited are invented.
Educational material. Not advice on any investment, tax, budget or market position.