Concentration Risk vs Wrong Way Risk: Two Different Faults

Concentration risk and wrong way risk are two separate faults, not one idea wearing two names. Concentration is about how a book is shaped: too much piled into one borrower, one connected set of borrowers or one industry, and it is already true this morning. Wrong way risk is about a link. The amount the bank is owed swells in precisely the state that leaves the borrower least able to settle.

Two things go wrong in a lending book and they sound alike when somebody says them quickly. The first is that there is too much of something. The second is that the amount at stake gets bigger at the worst possible moment. Both get described in meetings as too much risk in one place, and that sentence is the reason the two keep getting filed together, treated with the same remedy, and reported in the same table until one of them stops being reported at all.

Both faults appear at household scale first. Each is familiar there, and at household scale the two are rarely separated. A household running entirely on one salary is the first fault. Nothing has to happen for that to be true. The fault is true on a good Tuesday, with the salary credited and the fridge full. Now take a different household: one earner, paid in rupees, with a child studying abroad and a fee bill payable in dollars. If the rupee weakens, the fee bill in rupees gets bigger at the same moment the earner's purchasing power gets smaller. The dollar fee bill is the second fault, and bill and weakness arrive together rather than one after the other.

Nobody in the second household is doing anything reckless. The salary is fine. The fee is affordable today. The fault is not in either fact; it is in the line between them, and no examination of the salary alone or the fee alone will ever find it. The first fault is a property of the shape of the thing, and the second is a property of a relationship between two things that each look perfectly ordinary on their own.

The worked case is Vindhya Commercial Bank Limited, an invented mid-sized Indian commercial bank, and its ten largest single-name exposures, numbered C1 to C10, with every figure in Rs crore. How an exposure is built from what is drawn, what is committed and undrawn, and what a derivative position is currently worth is covered separately. Standing back from one exposure to look at all ten at once is what turns up a fault that no measurement of any single exposure can find.

Concentration is defined in full first, on this bank's own figures, with nothing said about the other fault. Then wrong way risk is defined in full, on the same bank's figures, with nothing said about the first. Only after both have been built separately do the two get laid against each other, on six criteria, and then measured against one common figure so their sizes can be compared honestly. Contrasting two things before defining either is how the two collapse into one blurred idea.

What is concentration risk, before any comparison at all?

Concentration riskToo much exposure in one name, one group, one sector or one anything, which is a fact about the shape of a book today. is the risk that too much of a book sits in one place. A single failure then takes an amount out of the bank that the bank cannot comfortably absorb. Three words in that sentence do the work. Take too much first. No absolute quantity is wrong on its own, and the judgement is always against a capacity. Then one place. The place can be a name, a set of connected names, a sector, a geography, a currency, a maturity or a source of funding. Then sits. The exposure is already there. Nobody has to do anything for the fault to exist.

The word sits is what most readers underrate, so hold on to it. Concentration is a fact about the book as it stands this morning, and it does not wait for any event to become true. If a bank has a quarter of its lending in one sector, that is true on the calmest day of the year. The failure that would hurt has not happened and may never happen, but the concentration is not a forecast of the failure; it is a measurement of what the book would lose if the failure came. Concentration can be computed with a list and a calculator, and no view about the world is needed at all.

So how is it found? By counting. Every exposure is taken, a decision is made about what to group it by, the groups are added up, and each total is compared with a limit. There is nothing clever in it, and that plainness is the point: any person holding the exposure list can reproduce the answer, and two people doing it independently get the same number. The hard parts of concentration are entirely in the definitions. The definitions are set out below, and they are where the fault hides.

Concentration at this bank, at three levels

Vindhya Commercial Bank Limited measures concentration at three levels and the case gives all three. At the level of one name, limit L1 caps exposure to a single counterparty at Rs 2,640 crore, and the largest name in the book, C1, Nirjhar Industries Limited, an invented steel and alloys maker, carries a reported total of Rs 2,496 crore. C1 runs at 94.5 per cent of the single name limitA cap on exposure to one counterparty, here limit L1, and this invented bank's own figure rather than a published requirement. and it is within.

At the level of connected names, limit L2 caps exposure to one borrower groupA parent and the entities it controls, treated as one exposure because they tend to fail together. at Rs 3,960 crore. Nirjhar Alloys Private Limited, a subsidiary wholly held by C1, carries a total of Rs 672 crore of its own. Add the two and the exposure to the Nirjhar group is 2,496 plus 672, being Rs 3,168 crore. The group total runs at 80.0 per cent of limit L2 and is also within.

At the level of a sector, limit L3 caps any one sector at Rs 7,056 crore, being 12.0 per cent of this bank's gross advances of Rs 58,800 crore. The infrastructure and power sector carries Rs 7,644 crore, being 13.0 per cent of gross advances and 108.3 per cent of the limit. The sector total is over, it is recorded as breach B1, it was first crossed in month 5, and at month 12 it is still open. It is the largest of this bank's three live limit breaches and it is the reason concentration is not an abstraction here.

Notice what the three levels do to each other. Read only the single-name level and the book looks tight but fine. Add the group level and a further Rs 672 crore appears that the first level could not see. Add the sector level and a figure three times larger than the largest single name comes into view, and that one is over its cap. Each level catches something the level below it is structurally unable to see. A bank that measures concentration at one level only has not measured concentration.

Two of those three rows are comfortable and the third is a live breachA limit crossed and still crossed, here breach B1 on limit L3, open since month 5 and accepted with a dated plan.. And the whole of that assessment was produced by adding numbers in a list and dividing by three caps somebody set in advance. No scenario was used, no view about interest rates or the rupee was needed, and nobody had to imagine a future. The signature of concentration risk is arithmetic on what is already there.

What is wrong way risk, before any comparison at all?

Wrong way riskExposure that grows in exactly the state of the world that makes the counterparty less able to pay. is the risk that the amount a counterparty owes the bank grows in exactly the state of the world that makes that counterparty less able to pay. The whole idea sits in the word exactly. The fault is not that a bad thing happens to both. One single movement in the world does both jobs at once. The same movement pushes the number up and pushes the payer down.

The opposite arrangement has a name too, and knowing it makes the first one easier to see. Right way riskThe opposite relationship, where the exposure shrinks in the state that weakens the counterparty, which is the case nobody worries about. is where the exposure shrinks in the state that weakens the counterparty. A bank that is owed more by a gold miner when gold prices rise, and less when they fall, has a position that gets smaller exactly when the miner is under pressure. Nobody loses sleep over that one. Same two ingredients, opposite sign on the link between them, completely different night's sleep.

So wrong way risk is not a quantity. Wrong way risk is the sign of a relationship, and a relationship needs two things and a state of the world before it exists at all. That is why it cannot be read off a list of exposures however carefully the list is drawn: a list has one row per counterparty, and this fault lives in a line drawn between a row and something that is not on the list.

Wrong way risk at this bank, at counterparty C7

Counterparty C7 is Wainganga Textiles Limited, an invented textile maker, and it is an importerA business that buys in a foreign currency, whose costs rise when its own currency weakens.. C7 sits at internal grade 6, it carries funded exposure of Rs 840 crore and a total reported exposure of Rs 1,056 crore, and it holds a US dollar forward against the bank whose current exposure is Rs 36 crore. Each part is unremarkable and the fault is not in any of them, so take the parts one at a time.

The borrower first. A textile maker at grade 6 with Rs 840 crore of funded exposure is the seventh largest name on this bank's published list and the second weakest grade in the top ten. C7 is not the biggest exposure, it is not the worst rated, and nothing about it would stop a reviewer's eye. The Rs 840 crore here is counterparty C7's funded exposure and nothing else. The same case also carries a Rs 840 crore economic value of equity change on limit L8, a different object entirely that belongs to the interest rate work.

The contract second. The definition of a forward belongs to the derivatives subject area, so the forward is named here and taken as a settled object rather than explained. One property of the forward matters. The contract is worth more to the bank when the rupee weakens against the US dollar. Today it is worth Rs 36 crore to the bank, and that Rs 36 crore is C7's derivative current exposure specifically. The same Rs 36 crore carries three other meanings in this case, including the negative leg inside C1's netting pair and the excess in the month 9 foreign exchange breach B2. Four different objects wear one number, and Rs 36 crore means nothing until the object behind it is named.

Now draw the line between them. A weaker rupee makes every consignment C7 imports more expensive in the currency C7 earns in. So a weaker rupee raises C7's import bill, squeezes its margin and weakens its ability to pay anybody, including this bank. And the same weaker rupee is what makes the forward worth more to the bank. The exposure grows in the same state of the world that weakens the counterparty, and that sentence is the entire definition of the fault, written on one real position.

The size of the position today is Rs 36 crore against C7's total of Rs 1,056 crore, being 3.4 per cent of one counterparty, and against the top ten total exposure of Rs 13,068 crore it is 0.3 per cent. On any measure, it is nothing. The smallness of the position is the single most important fact about wrong way risk, and it comes back below with force.

So what actually separates the two faults?

Concentration and wrong way risk each stand on their own now. Lay them against each other on six criteria. The second criterion sorts all the others, so take that one slowly.

Criterion WC1 asks what kind of statement each one is. Concentration is a statement about a shape. Wrong way risk is a statement about a dependence. A shape and a dependence are different categories of thing, in the way that a person's height and a person's allergy are different categories of thing, and no amount of measuring one yields the other.

Criterion WC2 is the one that decides the rest. Does the fault need something to happen before it is a problem? For concentration the answer is no, for the reason already given: Rs 7,644 crore sits in one sector this morning and breach B1 has been open since month 5 with no event of any kind required. For wrong way risk the answer is yes, and the state is named: a weakening rupee. Today the fault at C7 costs the bank exactly nothing, and that is not a defence of it, it is the definition of it.

Criterion WC3 asks how the search is run, and the two searches have nothing in common. Finding concentration is a matter of ranking and summing. The search takes the exposure list, a decision about what to group by, and the caps. Concentration is clerical work in the best sense, being reproducible, checkable, and equally available to anybody holding the same file. Finding wrong way risk takes two questions in sequence, and neither of them is answerable from the list. First: what state of the world makes this exposure bigger? Second: does that same state make this counterparty less able to pay? If the answer to both is the same event, the fault is there. If the second answer is no, the arrangement is right way risk instead, and nothing to worry about.

The second search has a shape worth keeping. The search starts at the contract and walks outwards to the world, then walks back in to the borrower from a different direction. Concentration never leaves the file. Wrong way risk cannot be completed inside the file at all. A bank whose whole assessment machinery is built on reports will produce a beautiful concentration pack every month and never once record this fault.

Recording it has a fixed shape too, and it is worth stating because at this bank the record is incomplete in a way that matters. A usable entry has four parts: the counterparty, the state of the world that grows the exposure, the reason that same state weakens the counterparty, and what the exposure actually becomes in that state. Three of those four are on file for C7. The fourth has never been computed. An entry missing the fourth part holds nothing anybody can disagree with, so nobody can review it.

Criterion WC4 asks what each fault is measured against, and here the asymmetry becomes uncomfortable. Concentration is measured against a limit, in rupees, and this bank has three of them on this subject: L1, L2 and L3. A limit is a good instrument for a quantity that already exists. Wrong way risk has no quantity to cap, so it is measured against a scenario instead. A slope cannot be capped. The only available question is what the exposure becomes when the named state arrives, and that question needs the state described and the position sized in it.

Criterion WC5 asks what makes each fault invisible, and the two answers are as different as everything else. Concentration goes missing inside definitions: which column the list was sorted on and where somebody drew the edge of a group. Both of those are addressed in the next section, on this bank's own list, and the result is startling. Wrong way risk goes missing for a completely different reason. Both halves of it are unremarkable when read alone, so the fault survives every review that examines the borrower and the trade one at a time. Nearly every review there is works that way.

Criterion WC6 asks what reduces each fault, and this is where getting the classification wrong stops being an intellectual matter and starts costing something. Because the two are reduced by unrelated actions, a fault filed under the wrong heading receives a remedy that does not touch it, and the remedy will be reported as complete.

Why does the list of the ten largest exposures depend on who drew it?

Concentration was described above as arithmetic on what is already there, and that is true. The arithmetic sits on top of two decisions somebody made before the counting began, and both decisions change the answer. The first is the ranking basisThe column a list is sorted on, which changes the list, which is why a top ten with no basis named is not a list of anything.. The second is where the edge of a group was drawn. Take them in that order, on this bank's own ten names.

Vindhya Commercial Bank Limited publishes its ten largest single-name exposures ranked by funded exposure, meaning the money actually drawn. On that basis the order runs C1 at Rs 1,680 crore down to C10 at Rs 540 crore. Now rank the same ten names on the same day by total exposure, meaning drawn plus the undrawn committed line plus what the derivative positions are currently worth. Nothing about the book has changed. Not one rupee has moved. And three of the ten change place while a fourth leaves the list altogether.

C6, Manjeera Housing Finance Limited, carries a Rs 300 crore undrawn line against C5's Rs 120 crore, so C6 rises from sixth to fifth. C5 falls from fifth to sixth for the same reason read the other way. C10, Betwa Speciality Chemicals Limited, rises from tenth to ninth, carrying Rs 180 crore undrawn and Rs 60 crore of derivative current exposure against C9's Rs 60 crore undrawn and none at all. And on the total basis a name the published list does not carry pushes past C9, Lohit Valley Tea Estates Limited. C9 leaves the ten altogether.

The name is Nirjhar Alloys Private Limited, the subsidiary of C1 met earlier, whose total exposure of Rs 672 crore would place it tenth. On the funded basis it would place ninth, at Rs 660 crore, and push C10 out instead. So the subsidiary is ninth on the funded basis, tenth on the total basis, and nowhere at all on the published list. The bank reports it inside the group line rather than as a single name. Note the trap in those two figures: Rs 660 crore is Nirjhar Alloys' funded exposure and it is also C9's total exposure, two different objects wearing one number in the same case.

Now the second decision, the edge of a group, and it is the larger of the two. Add C1's Rs 2,496 crore to Nirjhar Alloys' Rs 672 crore and the exposure to one borrower group is Rs 3,168 crore. The Rs 3,168 crore does not appear on the list above. The group total does not appear on the funded version of the list, or the total version, or in any cell of the table. The group exposure exists only once somebody connects the parent and the subsidiary, and forcing exactly that connection is what limit L2 was created for. Against L2's Rs 3,960 crore it runs at 80.0 per cent, which is comfortable, and it would have been invisible without the definition.

Put the two decisions together and the shape of the concentration problem changes. The difficulty is not that the counting is hard. The counting is trivial. The counting sits on a sort column and a group definition, both chosen upstream, and anybody handed the output without being told either has been handed a number that cannot be checked. A list of ten largest exposures with no basis stated is not a list of the ten largest anything.

How big is each fault, measured on one common figure?

Two faults measured in different units cannot be compared. Concentration has been in rupees against a cap. Wrong way risk has been in dependence, and a dependence has no unit at all. So put them both into rupees, and do it in the only way that is honest: measure each one by how far it is from where it should be.

For concentration that distance is already on record. The infrastructure and power sector carries Rs 7,644 crore against limit L3's Rs 7,056 crore, so it is over by 7,644 less 7,056, being Rs 588 crore. Nothing has to happen for that Rs 588 crore to exist. The excess existed last month. Rs 7,644 crore in this case is also the sum of grade 1 and grade 2 exposures, being 2,940 plus 4,704, so the object needs care. The best-rated two grades and the sector in breach share a number and nothing else.

For wrong way risk the level today is not the fault, so the distance is a growth rather than a level. C7's derivative current exposure is Rs 36 crore. If a weaker rupee multiplies it by some figure, call the figure w, the growth in it is 36 times w less 1. Setting that growth equal to Rs 588 crore gives 36 times w less 1 equals 588, so w less 1 is 16.33 and w is 17.33. The contract exposure at C7 would have to grow more than seventeen times before its growth reached the amount by which one sector is already over its cap this morning.

Commit to an answer before the control below moves. Expectation and arithmetic sit a long way apart on this one.

Where does the distinction between the two actually break down?

The distinction breaks in one specific place, and it breaks quietly. Both faults sound like too much of something, so somebody classifies wrong way risk as a kind of concentration. Once it is filed there, it receives what everything in that file receives: a limit, a utilisation percentage, and a row in the monthly pack.

When does the difference between them change a decision?

Four people read the same two faults and act on them in four unrelated ways. Watching the four is the quickest test of whether the distinction is real or academic.

The credit officer inside the bank has to choose an action, and the choice is decided entirely by the classification. On the sector she can act today. The Rs 588 crore excess is a quantity, quantities can be sold, syndicated or allowed to run off, and breach B1 already carries an accepted remediation plan running to month 18 with a named risk owner in the governance record. On C7 there is nothing to sell that would help. The only actions that touch a dependence are changing the contract, taking security that holds its value in the state that hurts, or pricing the dependence, and none of them looks like reducing an exposure.

The analyst outside the bank cannot recompute any of this, and does not need to. Three questions cost nothing and separate a bank that has thought about the problem from one that has not. On what basis is the top ten ranked? A list with no basis is not a list of anything. How is a borrower group defined and where was the edge drawn? The Rs 3,168 crore at Nirjhar exists only after that definition is applied. And are any exposures recorded as wrong way, with the state of the world named. A pack with three concentration tables and no answer to the third question has measured the fault that is easy to measure.

The board committee reads the two as different kinds of business. A limit breach is a decision to be made: accept it with a dated plan, as this bank has done with B1, or act on it. A dependence with no measurement is a gap in the machinery, and the correct instruction is not a limit but a body of work: name the scenario, get the notional, form the credit view.

The mechanism is identical at every scale, so take the household version too. A person with all their savings in one employer's shares carries the first fault, and it is true today whether or not that employer ever stumbles. A person whose salary comes from that same employer as well carries the second fault. The state of the world that destroys the savings is the same state that ends the income. The first is fixed by selling some shares; the second is not fixed by selling shares at all, and telling the two apart is what decides whether the remedy does anything.

What happens when a concentration and a severe state arrive together?

One more figure, and it is handed forward rather than worked here. Vindhya Commercial Bank Limited's own loss capacity is Rs 2,400 crore, being the loss that reaches its own capital floor, and that floor is the bank's own figure rather than anybody's requirement. 6,000 times 0.40 is 2,400, so at its assumed 40.0 per cent loss given default a loss of Rs 2,400 crore needs Rs 6,000 crore of exposure to default. And Rs 6,000 crore is 78.5 per cent of the infrastructure and power sector's Rs 7,644 crore, being the sector that is already in breach.

Read that slowly. The entire path from a comfortable book to the bank's own capital floor lies inside one sector that has been over its cap since month 5. When a state of the world does finally arrive, the fault that needed nothing to happen becomes the channel through which the loss arrives. Working backwards from a loss to the state that produces it is a reverse stress test, and the method for it is set out under reverse stress testing, alongside the full step-by-step assessment of concentration. Note the object once more before leaving it. The Rs 2,400 crore here is this bank's loss capacity, and the same figure appears in this case as tier 2 subordinated bonds and as a liquidity bucket gap, two different things entirely.

Where do the rules on large exposures and wrong way risk come from?

The order of the sources matters. The idea of capping exposure to a single counterparty and to connected counterparties, and the treatment of wrong way risk as a named category inside the counterparty credit risk framework, originate with the Basel Committee on Banking Supervision at the Bank for International Settlements, bis.org. The Basel Committee is where the reasoning is set out and where the definitions were first written down.

A global standard is not what binds an Indian bank, and naming only the first source is a confident and common error. What actually binds on large exposures, on how a borrower group is defined for that purpose, and on sector exposure comes from the Reserve Bank of India at rbi.org.in. Name the standard, then name what binds, then send the reader to the text of both. State no figure from either.