Counterparty Risk: What Clearing Moves Rather Than Ends

Counterparty risk is the risk that the party who owes does not pay. Putting a clearing corporation between the two sides does not delete that risk. Clearing relocates the risk, away from one named stranger and onto the arrangement itself. Collateral is what answers the relocated version: something goes up before the promise is carried, gets called again while the position is still open, and is never allowed to fall behind what is owed.

Every promise in this machinery is one party's asset and another party's obligation, and an obligation is worth exactly what stands behind it. Whether the obligation is worth that much is a question about somebody's credit standingHow good a party's promise to pay is judged to be. It is slow to move, awkward to measure, and nobody can recheck it every evening., and credit is a slow, blurred, expensive thing to measure. Credit cannot be re-examined at every close of business. So the arrangement quietly swaps the question for a different one, chosen because it can be answered in a moment: what has this party already handed over, and did today's call arrive.

What is counterparty risk, in the plainest terms?

Counterparty risk is the risk that whoever owes does not pay, and it has nothing whatsoever to do with whether the holder was right about the price. Failing to pay and being wrong about the price are two different failures with two different cures, and the language keeps them apart for exactly that reason. A position holder can read the market perfectly, watch the settlement price move in their favour, be owed money on the evening it happens, and receive nothing, because the party on the other end has stopped existing in any way that matters.

Take it out of the market first. A household has let out the upstairs room and is owed six months of rent by a tenant who has moved out one night without saying where. The household did not misjudge the rent. The rent was fair, the room was worth it, and every rupee of that judgement still stands. The person failed, not the price. Nobody in that household would say they got the rent wrong; they would say the tenant did not pay. The tenant failed, the rent did not, and clearing is a machine built around that one distinction.

Now put it back into a position. A holder stands on one unit of an invented reference asset, and that asset hands its holder nothing at all over the period it is held. The spot price is Rs 2,000.00/-. The same figure read a second way is the exposureHow much of the referenced thing a position is riding on. It is a size, not a sum anybody has handed over. on one unit. Suppose the settlement price for the day comes in at Rs 2,080.00/-, a 4.0 per cent move in the holder's favour, so a PAYMENT of Rs 80.00/- is owed to the holder at that close. Market riskThe chance that the price itself moves against the position. A completely separate question from whether anybody pays what they owe. has gone the holder's way. Counterparty risk is a question nobody has asked yet: does that Rs 80.00/- arrive.

Read side by side, the difference between the two panels is not a matter of degree. In the left panel the machinery is doing precisely what it was built to do: the settlement price moved, a PAYMENT of Rs 80.00/- was worked out, and it left the holder's account. In the right panel the machinery worked out an identical Rs 80.00/- in the holder's favour, and then nothing happened. Every part of the arithmetic was correct. The arithmetic was never the problem.

What does putting a clearing corporation between the two sides actually change?

Two people who agree a price directly are stuck with each other. Each one now carries a promise from a specific party whose accounts they have not seen, whose other obligations they know nothing about, and whose evening they cannot observe. Multiplied across every pair of parties dealing in the same contract, that gives a web where everybody has to form a private opinion about everybody else.

Once a position is cleared, the two original sides stop facing each other altogether: each faces a clearing memberThe party that stands in front of the clearing corporation on a position holder's behalf, and answers for what that position holder owes., and each clearing member faces the clearing corporation. The stranger the price was agreed with is now, from the position holder's side of the arrangement, irrelevant. Their credit is not what is being carried. No opinion about them is needed, and a useful one could not be formed anyway. Removing the need for that opinion is the first half of what clearing changes, and it is genuinely large.

Work the count yourself. The count is the whole argument. Among 5 parties dealing straight with one another there are 5 times 4 over two, or 10 separate two-party promises, and each of the five parties is carrying 4 of them. Put one party in the middle and there are 5 promises in total, one per party, and each party is carrying exactly 1. Nothing about the underlying positions changed. The count of relationships anybody has to form a view about fell from four to one.

Here is the half that gets dropped, and dropping it is where the misunderstanding starts. Somebody still has to pay. The obligation did not evaporate when it was rerouted; it acquired a new holder. The credit now being carried is that of a clearing member and, behind the member, of the arrangement standing in the middle. A clearing member's credit is a much better thing to be carrying than the credit of a stranger met over a price, and it is not nothing. The question moved. The question did not close.

Drag it down to 2 and something honest happens that most explanations of this quietly skip. With two parties dealing directly there is exactly 1 promise; put one party in the middle and there are 2. At 3 parties the two counts are equal at 3 apiece. Standing somebody in the middle only starts to reduce the count of promises from 4 parties upward, and by 8 parties it is 28 against 8. The middle is not a tidying device that works at every size. The middle is a structure whose whole value is that the direct count grows with the square of the number of parties while the spoke count grows in a straight line, and at small numbers that advantage has not appeared yet.

How does collateral answer a question that is really about credit?

There is now a clearing member on the other end instead of a stranger. A clearing member is a better party to be carrying than a stranger. Being better does not establish whether the member will pay tomorrow, and here the arrangement makes the move that everything else rests on.

Instead of asking whether a party is able to pay, which no one can verify at every close, the arrangement asks what that party has already handed over, which anyone can verify in seconds. Notice what has been done there. The hard question was not answered. The hard question was set aside in favour of a question of a completely different kind: not a judgement about a party's condition, but a balance sitting in an account, with a number on it, that either covers what is owed or does not. A judgement takes an analyst a week and can still be wrong. A balance takes one look.

Here is what is actually posted. Two figures come out of it, and taking only the first understates the position while taking only the second leaves no idea where it came from. Initial marginWhat a holder lodges at the outset, ahead of anything going wrong, simply for permission to carry the position. of 8.0 per cent of the Rs 2,000.00/- exposure comes to Rs 160.00/- on one unit. The Rs 160.00/- is the first half. The second half is the ratio hiding inside it. The Rs 2,000.00/- riding on the position divided by the Rs 160.00/- lodged behind it gives embedded leverageHow many rupees of exposure one rupee of posted collateral is carrying. Here it is twelve and a half of them. of 12.50 times. And a 4.0 per cent adverse move on the reference asset is Rs 80.00/-, or 50.0 per cent of everything that was put up.

The 50.0 per cent is worth reading against the alternative. Held outright, the reference asset takes a Rs 80.00/- swing as four rupees in every hundred, hardly worth remarking on. Carried on Rs 160.00/- of lodged collateral, that identical Rs 80.00/- is half of everything handed over. The movement did not change at all. The base it gets measured against changed, and that is why collateral has to be watched rather than simply collected.

Now the label that matters more than the arithmetic. The 8.0 per cent is a teaching figure, put there so the ratio can be seen at all, and it is not a requirement that anybody has to meet. The Securities and Exchange Board of India (SEBI) fixes what is genuinely called for, and publishes at sebi.gov.in; the real figure differs by contract, differs by day, and moves. The shape survives whichever number turns out to be the live one: a small posted amount stands under a much larger exposure, and a modest move in the reference asset then eats a large fraction of the collateral. Change the percentage and the ratio changes with it; the fact that there is a ratio worth watching does not.

Why does collecting every day stop the exposure from growing?

Collateral posted at the start is a stock. The daily call is a flow, and the flow is where this arrangement is cleverer than it looks.

The call comes before the loss does. Because the day's movement is collected at the close, the amount anybody owes at any given moment is only what has happened since the last call, rather than everything that has happened since the position was opened. The obligation is emptied and refilled every day. The obligation never gets the chance to accumulate into the kind of number that somebody might reasonably decide to walk away from.

The consequence is not really about arithmetic at all; it is about what anybody is tempted to do. A promise that is emptied every evening never grows to the size that would make it worth breaking. An obligation of one day's movement is an inconvenience. An obligation of a year's accumulated movement is a decision. The daily call is what keeps every obligation in the first category, and it does so by asking for money at the one moment when the money can still be found. The money can still be found while the position is still open and nothing has gone wrong yet.

What is still uncovered when every part of this works properly?

The answer is not nothing. Two gaps survive even when every component behaves exactly as designed. Both of them are ordinary consequences of the fact that collection happens at moments rather than continuously.

The first gap is the movement since the last call. The movement has already happened. Because the moment for collecting it has not arrived, the movement has not yet been collected. On the figures worked here, a 4.0 per cent move is Rs 80.00/- a unit, or 50.0 per cent of the Rs 160.00/- that was put up, and for as long as that gap is open the arrangement is carrying it.

The second gap opens after somebody stops answering. A position belonging to a party that has gone quiet has to be closed outEnding a position without the holder of it choosing that moment. Somebody else decides, because the holder is no longer answering., and closing takes time. During that time the price does what prices do. The length of that time, and the size of the move inside it, depend on liquidityHow quickly something can be turned into cash without shifting its own price on the way out. and on conditions nobody controls.

The initial margin is there for those two gaps, and that is why the movement it answers to is one still lying ahead rather than one already on the record. The claim is a strange one until what it is doing becomes clear. Variation collection deals with what already occurred; there is a figure for it and it gets settled. Initial margin is the buffer for the part nobody has observed: the movement in the window before the next call lands, and the movement while a position is being unwound by somebody other than the party that entered it. A buffer cannot be sized against history when the thing being buffered has not occurred.

What happens when a clearing member fails, and why is the answer an order?

Somewhere behind everything above sits the case this whole structure exists for: a clearing member does not meet what it owes. Not a price move, not a late payment that turns up in the morning, but a party that cannot cover its obligation at all.

A sequence was decided in advance: the losses get met from resources reached for in a stated order, and that order was written down and published long before anybody needed it. The contents of that order are settled and published by the authority.

The steps of that sequence, and the amounts inside it, belong to the authority. SEBI settles every part of it, at sebi.gov.in. The sizes of the resources change. The order in which they are reached for was fixed in advance and made public, and that is the part a reader is entitled to know. Nobody then works out on the night who bears what.

Why is a sequence fixed in advance worth more than a larger pile?

The argument for why the sequence is the thing worth having has nothing to do with size.

A settled sequence cannot be negotiated at the moment of failure. The moment of failure is precisely when nobody has time to negotiate, when every party's interest points in a different direction, and when the people who would have to agree are the people with the most to lose from agreeing. An arrangement that has not decided by then will decide badly, or will not decide at all.

A settled sequence also tells every participant where they stand before they join, rather than after something has happened to them. Somebody weighing up whether to place collateral with an arrangement can read what happens to it, in what order, under conditions that have not arisen. Read after a failure, the same document is read by a claimant. Read before, it is read by a chooser, and a chooser is in a completely different position.

An outsider who was nowhere near the decision can also go and audit it. A choice made on the night cannot be audited that way at all. The only record of such a choice is whatever the people who made it say afterwards. Fix the rules first, at a moment when there is still no way of telling whose side they will land on, and they hold when the pressure arrives. Fix them afterwards and they get argued about instead. The published order is one instance of that general shape, and for that reason it counts as the protection rather than as a description of one.

What may be concluded from all this, and what may not?

A careful reader can go wrong here in a way that feels like care. The structural fact is real: an order exists, it was settled in advance, it is published. Converting the structural fact into a statement about safety is tempting, and the conversion does not go through.

The shape of the answer can be stated. The answer itself cannot be. A design is a different kind of fact from an outcome, and no quantity of the first ever delivers the second. Knowing that an arrangement has a settled sequence establishes something about how it was designed and precisely nothing about whether it is sufficient.

How does the whole thing look worked through on one unit?

Everything above, on one unit of the invented reference asset, with each figure labelled for what kind of thing it is. Every figure is assumed rather than read off a market, and a live quote would move all of them together. Because the reference asset returns not one paisa to whoever holds it during the period, no payout appears anywhere in the arithmetic.

And the structure, counted rather than asserted. Among 5 parties dealing directly, 5 times 4 over two gives 10 two-party promises, and each party is carrying 4 of them. With one party in the middle the count is 5 and each party is carrying 1. Push it to 8 parties and the direct count reaches 28 against 8 spokes. Pull it back to 2 and the direct count is 1 against 2 spokes. At that setting the middle is not yet earning anything.

Where does the collateral side of this question end?

Every question above has been about one side of the arrangement: what a position holder puts up, when it gets called, and what happens to it afterwards. There is another side, and it is a large subject in its own right.

How a clearing corporation is paid for, what it holds behind that published sequence, who contributed those resources, who governs it, and what makes it a separate body from the exchange rather than a department inside one are all worked out properly somewhere else. The collateral side of the question is worked above, and the funding side is covered separately.

So am I protected?

All of that arrives at the one thing that actually matters to a reader, and it is fair to ask for it directly.

No honest answer of yes can be given at this distance. Whether a particular position is protected depends on what was lodged against it, where that collateral is sitting tonight, what the published sequence actually contains, and whether any part of it has ever had to carry a real loss. SEBI decides most of that, at sebi.gov.in. Understanding how a machine works has never been a reason to stand inside it.

Something smaller and more useful is available instead. A particular arrangement's requirements are set out by the authority rather than in a teaching text. The published sequence is read at its source. Of the four fields in the practitioner block above, the ones that can be filled in straight away are worth marking, and every blank one is a question rather than a detail. None of that is a substitute for the answer wanted, but it is what exists in place of one.

The reference asset priced at Rs 2,000.00/- and the position holder standing on one unit of it are invented.
Educational material. Not advice on any investment, tax, budget or market position.