Liquidity Coverage Ratio: Surviving Thirty Days of Stress

The liquidity coverage ratio sets a stock of monetisable assets against thirty days of stressed net cash outflows. At Vindhya Commercial Bank Limited, invented, a buffer of Rs 14,400 crore over modelled net outflows of Rs 11,520 crore reads 125.0 per cent. Nobody watched that outflow happen. Every run-off and inflow factor inside it is the bank's own working number, not a requirement.

Two things have to be true before that sentence means anything. The buffer has to be a real stock, sitting there, countable this morning. The denominator does not have to be real at all, and it is not. The thirty days it describes have not happened, so nobody measured a thirty day net cash outflow of Rs 11,520 crore at Vindhya Commercial Bank Limited. Somebody chose four categories of money that could leave, chose a factor for each, chose two categories of money that might arrive, refused to count all of one of them, and subtracted. Six choices. The whole denominator is those six choices, and the rest of this guide follows them one at a time.

What does the liquidity coverage ratio actually measure?

The ratio measures one stock against one flow, and only one of the two exists. The stock is the buffer: assets the bank believes it could turn into cash quickly and at close to their value, held so that a bad month can be paid for out of what is already on the premises. The flow is the money the bank assumes would leave over thirty days if that bad month arrived. The stock is counted and the flow is modelled, and confusing those two is the single most common error a reader makes with this measure. The first divided by the second, expressed as a percentage, is the reading.

The household version has an identical shape and is much easier to feel. A household keeps money aside for a bad month. Knowing whether the money is enough means deciding what a bad month costs. The bad month has not arrived, so no bank statement anywhere states that figure. So the household makes a list: rent will still be due in full, groceries will still be due in full, the school fee will still be due, and the freelance payment that usually lands on the fifteenth might not come at all. The list is a denominator built out of assumptions. Somebody who assumed the freelance payment always lands would get a much friendlier answer from the same money in the same drawer. The drawer did not change; the assumptions did, and the ratio moved anyway.

At Vindhya Commercial Bank Limited six such assumptions carry the whole denominator. Four are run-off factorsThe share of a balance assumed to leave inside the thirty day window, applied to a category rather than to a named customer., applied to four categories of funding, and two are inflow factorsThe share of an expected receipt allowed to be counted inside the window, which is capped so the measure does not lean on money arriving., applied to two categories of expected receipts. Every one of the six is the bank's own working number, and none of them is a requirement, a published rate or a fact about banking in India. Read the drawing below as a fraction with a footnote: the top half was counted, the bottom half was decided.

Every factor in the panel on the right is a working number of the invented bank, not a published rate, an official minimum or a date of force. Published rates and minimums exist, they matter enormously, and the bodies that issue them are named below. A reader has to be able to see which of the two a factor is before the headline percentage is worth reading at all.

What goes into the buffer, and why is part of the cash line missing from it?

The numerator is a stock of high quality liquid assetsAssets an institution counts in its buffer because they can be monetised quickly at close to their value, on a definition set by an authority rather than by the institution itself., and the word doing the work in that phrase is not liquid but counted. An asset qualifies because a rule says it qualifies, at a value the same rule sets, and an institution that thinks its own holdings are wonderfully saleable does not get to promote them on that basis. A buffer is a claim about availability on the day, not a claim about asset type. At Vindhya Commercial Bank Limited the buffer has four components, numbered H1 to H4 in the bank's own working papers, and they add to Rs 14,400 crore.

The shares come before the rupees. Buffer components H2 and H3, the two lines of government securities, are Rs 11,520 crore between them, or 80.0 per cent of the buffer of Rs 14,400 crore. The concentration is not an accident and it is not a criticism: the assets that hold their value best in a bad week are the assets everybody else also wants to hold in a bad week, and sovereign paper is the deepest market a bank in almost any country has access to. A repo, a certificate of deposit and a government security each behave differently, and each is a subject in its own right, taught in the fixed income material. In this guide they are names attached to figures.

Now the line that teaches the most, and it is the smallest one. The balance sheet of Vindhya Commercial Bank Limited shows Rs 4,800 crore of cash and balances with the central bank. Buffer component H1 is Rs 1,920 crore, or 40.0 per cent of that line. The other Rs 2,880 crore of the cash line is held against the reserve requirementBalances an institution has to keep with the central bank, which is why part of a cash line can sit on the balance sheet and still be unavailable. and therefore cannot be spent on a bad Tuesday. Cash that must stay where it is fails the buffer test however liquid it looks, and that single exclusion is the cleanest illustration in this guide of what a buffer actually claims.

How is the outflow side built, category by category and factor by factor?

One row at a time, and each row is the same three things: a balance, a factor, and their product. The entire method is those three things. The balance is real and comes off the balance sheet. The factor is a judgement about what share of that balance walks out inside thirty days of stress. The product is what the model says leaves. A category with a small balance and a savage factor can outweigh a category with a huge balance and a gentle one, and reading the rows is the only way to see it happen.

Vindhya Commercial Bank Limited uses four outflow categories. Retail deposits are savings accounts of Rs 26,400 crore plus retail term deposits of Rs 29,280 crore, being Rs 55,680 crore, and the bank applies its own assumed run-off of 7.5 per cent, giving Rs 4,176 crore. Unsecured wholesale fundingMoney from institutions and large depositors with no security given against it, which is assumed to leave faster than any other deposit category. is current accounts of Rs 9,600 crore plus wholesale term deposits of Rs 11,520 crore, being Rs 21,120 crore, at the bank's own 40.0 per cent, giving Rs 8,448 crore. Borrowings of Rs 8,400 crore run off at the bank's own 25.0 per cent, giving Rs 2,100 crore. Undrawn committed lines of Rs 12,000 crore are assumed to be drawn at the bank's own 10.0 per cent, giving Rs 1,200 crore. Add the four and gross outflowsThe sum of every category's balance times its run-off factor, before any inflow at all is subtracted. are Rs 15,924 crore.

The fourth row surprises people, and it deserves a sentence of its own. Undrawn committed lines are not money the bank has lent. Undrawn lines are promises to lend, sitting off the balance sheet, and in an ordinary month they cost nothing. In a stressed month a customer who fears that credit is about to become scarce draws down every line available before anybody can withdraw it. An undrawn line behaves like a deposit that runs the wrong way, and a measure that ignored it would miss the exact behaviour a stress produces. The everyday parallel is the overdraft facility on a small trader's account: unused for eleven months, and drawn to the last rupee in the twelfth, precisely when the bank can least afford it.

The bar column is where to start. The second bar is twice the first and it comes from a balance well under half its size. Nothing about that is hidden, and nothing about it is visible in a headline percentage either. The inversion becomes visible only when the worksheet is read one row at a time. Reading it that way is the argument for ever publishing a worksheet at all.

Why are inflows capped when outflows are not?

Because the stress the measure describes is happening to the bank's borrowers as well. Money the bank expects to receive inside the window is real, contractual and mostly reliable in an ordinary month, and an ordinary month is not what is being tested. So the inflow side is deliberately handicapped: some receipts are counted at less than their face value, and the count is capped so that a bank can never talk its denominator down to nothing by pointing at how much is due. Outflows are stressed upward and inflows are stressed downward, and that asymmetry is the design rather than a flaw in it.

Vindhya Commercial Bank Limited counts two inflow categories. Performing loan repayments falling due within thirty days are Rs 4,008 crore, and the bank's own inflow factor of 50.0 per cent counts Rs 2,004 crore of them. Money placed with another bank for a few days is the closest thing on the asset side to cash the institution has already decided to get back, so interbank placements of Rs 2,400 crore are counted at 100 per cent, giving Rs 2,400 crore. Counted inflows are therefore Rs 4,404 crore. Both factors are the bank's own working numbers and neither is a published rate.

Notice what the first of those two rows actually says. Rs 2,004 crore of contractual repayments, from borrowers who are paying today, is thrown away on purpose. A treasurer looking at the worksheet can point at it and say the money is coming. The discipline exists for exactly that reason. A measure that let an institution rely on its borrowers paying on time during a funding stress would have assumed away the thing it was built to test. The household version is the freelance payment again. The payment belongs in the plan for an ordinary month. Losing the job and losing the client have a way of arriving together, so the payment does not belong in the plan for the month the job is lost.

The arithmetic is now complete. Gross outflows of Rs 15,924 crore less counted inflows of Rs 4,404 crore give a thirty day net cash outflowGross outflows less counted inflows, which is the denominator of the ratio and a modelled figure rather than a measured one. of Rs 11,520 crore, and the buffer of Rs 14,400 crore over that denominator reads 125.0 per cent. One number to remember and one trap to avoid with it: the thirty day net cash outflow of Rs 11,520 crore and the government securities of Rs 11,520 crore inside the buffer are two different objects that happen to be the same size, and they sit on opposite sides of the same ratio. Which of the two is meant has to be said every single time.

Which category is a fifth of the balances and half the outflow?

Unsecured wholesale funding, and seeing that line is seeing what the whole measure exists to say. The four outflow categories sit on balances of Rs 97,200 crore between them. Retail deposits are Rs 55,680 crore of that, being 57.3 per cent, and they produce Rs 4,176 crore of outflow, being 26.2 per cent of gross outflows of Rs 15,924 crore. Unsecured wholesale funding is Rs 21,120 crore, being 21.7 per cent of the balances, and it produces Rs 8,448 crore, being 53.1 per cent of gross outflows. A fifth of the money produces over half the modelled outflow, and the inversion happens entirely inside the factor column.

The reason is not mysterious and it is worth saying in plain words. The bank's own assumed run-off on unsecured wholesale funding, at 40.0 per cent, is 5.33 times its assumed run-off on retail deposits, at 7.5 per cent. Behind that multiple is a view about behaviour. A treasury team at another institution watches a screen all day, moves at the first sign of trouble and has nothing tying it to the bank. A salaried household with a savings account watches nothing, moves slowly, and has a standing instruction and a decade of habit tying it to the branch. The factor column is a behavioural assumption wearing a percentage sign, and it is where the argument about this measure always lives.

Think of a small restaurant with two kinds of customer. Two hundred neighbourhood regulars come in for a hundred rupees each, and one office contract books lunch for eighty people every weekday. The regulars are the bigger revenue. One phone call ends the contract, and the regulars leave one funeral, one move and one bad meal at a time. The contract is the bigger risk. Any owner who ranked those two by revenue would rank the wrong one first, and any reader who ranks this bank's funding by balance does the same thing.

How far can any one assumption move before the ratio reaches 100.0 per cent?

There is only one distance, so start with it rather than with the assumptions. The buffer of Vindhya Commercial Bank Limited is Rs 14,400 crore and its thirty day net cash outflow is Rs 11,520 crore, so the net cash outflow can rise by Rs 2,880 crore before the two are equal. The gap of Rs 2,880 crore is the headroomThe amount by which the modelled outflow could rise before a ratio reached 100.0 per cent, which in this guide is one figure that every sensitivity turns out to be., and it is a single number. Every sensitivity in this measure is that one Rs 2,880 crore wearing a different coat.

Worked out one at a time, they give five sentences that sound unrelated and are not. Retail run-off rising from 7.5 to 12.67 per cent on the retail balance of Rs 55,680 crore adds Rs 2,880 crore. Unsecured wholesale run-off rising from 40.0 to 53.64 per cent on Rs 21,120 crore adds Rs 2,880 crore. Borrowings run-off rising from 25.0 to 59.29 per cent on Rs 8,400 crore adds Rs 2,880 crore. Undrawn line drawdown rising from 10.0 to 34.00 per cent on Rs 12,000 crore adds Rs 2,880 crore. Counted inflows falling from Rs 4,404 crore to Rs 1,524 crore, a fall of 65.4 per cent, takes away Rs 2,880 crore of subtraction. The same movement is simply seen from the other side.

One distance behind five rows is what makes a sensitivity table coherent instead of a list of disconnected facts. There is one distance and five ways of walking it, and once a reader sees that, no individual row in such a table can surprise them again. The same fact settles how any sensitivity table should be read: the headroom comes first, and each row is then checked for moving the denominator by exactly that amount. If one of them does not, either the row is wrong or it is measuring something other than the crossing.

Which assumption does the answer depend on most?

Ask that question and watch it split in two. There are two honest ways to measure how far an assumption has to move, and they give different winners. Measured in percentage points, retail run-off has to move least: 5.17 points, against 13.64 for unsecured wholesale, 24.00 for undrawn line drawdown and 34.29 for borrowings. On that ranking retail deposits are the most sensitive assumption in the computation. Measured as a multiple of the factor itself, unsecured wholesale has to move least: 1.34 times, against 1.69 for retail, 2.37 for borrowings and 3.40 for undrawn. On that ranking unsecured wholesale funding is the most sensitive. Both rankings are arithmetically correct and they name different categories at the top.

Neither one is a trick. In points, retail moves least because the balance underneath it is the largest on the table, so a small change in the factor moves a lot of rupees. As a multiple, wholesale moves least because its factor is already high, so a proportionally modest push on an already severe assumption produces the same rupees. Which assumption matters most is a fact about how the question was asked and not a fact about the bank. The basis question is worth carrying into every sensitivity table.

The practical consequence is unglamorous and important. A sensitivity table published without stating its basis has answered a question nobody can identify. A reader looking at a list headed most sensitive assumption cannot tell whether the author measured points or multiples, and so cannot tell whether the top row means the factor that would move most easily or the factor that sits on the most money. The basis has to be established before the order is read, and if nobody can name it, the ordering is not information.

Can any single early warning indicator break this ratio on its own?

No, and that answer is worth more than it looks. Vindhya Commercial Bank Limited keeps seven early warning indicators numbered W1 to W7 with its own amber and red triggers on each. W5 watches how much of the undrawn committed lines customers have actually drawn, and the bank's own triggers put it amber at 15.0 per cent and red at 25.0 per cent of the total. The drawdown assumption inside the coverage computation is 10.0 per cent, two thirds of the bank's own amber trigger. So run the indicator forward and see where the reading goes.

At W5 amber, drawings of Rs 1,800 crore take the thirty day net cash outflow to Rs 12,120 crore and the reading to 118.8 per cent. At W5 red, drawings of Rs 3,000 crore take the net cash outflow to Rs 13,320 crore and the reading to 108.1 per cent. Breaking the ratio on that assumption alone would need a drawdown of 34.0 per cent, so the indicator is red, the treasurer is in a difficult meeting, and the coverage reading is still comfortably above 100.0 per cent. A stock divided by a sum of four outflows and two inflows cannot be broken by any one of its parts moving alone, and that is arithmetic rather than an opinion about the bank.

Now hold the two facts together. The pair is the lesson. The ratio is robust to any single move and therefore slow. The indicators are sensitive to a single move and therefore noisy. An institution that watched only the ratio would learn nothing until several things had gone wrong at once; an institution that watched only the indicators would spend its year chasing amber lights that meant nothing on their own. Both are needed, and they are needed precisely because they fail in opposite directions. The everyday version is a car. The fuel gauge is slow, honest and useless as an early warning. The noise the engine makes is immediate, alarming and often nothing at all.

How does 125.0 per cent sit beside a scenario that consumes most of the same buffer?

The reconciliation comes before the second number. In the wrong order the two look like a contradiction, and they are not. Vindhya Commercial Bank Limited runs two different sets of outflow assumptions over the same thirty days. The computation in this guide produces a net cash outflow of Rs 11,520 crore, an average of Rs 384 crore a day. The bank's own severe scenario is a harsher set of assumptions it wrote for itself, and it produces Rs 13,200 crore over the same window, an average of Rs 440 crore a day. Divide one by the other: 13,200 over 11,520 is 1.146. The bank's own severe scenario is 14.6 per cent harsher over thirty days than the computation this guide has built, and the same buffer of Rs 14,400 crore therefore reads 109.1 per cent against it instead of 125.0 per cent.

Now the second number can be stated safely. Under that harsher scenario the buffer of Rs 14,400 crore has Rs 13,200 crore of it consumed by day 30, or 91.7 per cent, leaving Rs 1,200 crore or 8.3 per cent. One buffer measured against two different assumption sets gives both readings, so the headline says a quarter more than needed, the bank's own scenario says almost all of it gone, and both are correct. How long that remaining Rs 1,200 crore lasts, and what the resulting horizon in days actually means, is the subject of the survival horizon and is covered separately; it is named here only as the other reading of the same buffer.

Who actually reads this number, and what do they do with it?

Four different people read it and only one of them can see how it was built. The difference in what each can see decides what each of them can honestly say. Inside Vindhya Commercial Bank Limited the asset liability management committee G4 sets the behavioural assumptions, so its members are the only readers holding both the reading and the factor column. The committee never asks about the level of the reading. The real question is which factor it would have to defend if the reading fell, and the arithmetic answers that one: retail run-off sits on the largest balance, so it moves the denominator fastest per point of change.

An analyst outside the bank reads a published ratio and cannot see the factor column at all. The useful analytical move is therefore comparison rather than level. A bank funded mostly by retail deposits and a bank funded mostly by unsecured wholesale funding build wildly different denominators from the same balance sheet size. Two banks reporting the same reading with very different funding mixes are therefore not in the same position. The reading is a number; the funding mix underneath it is the information, and only one of the two is in the headline.

A lender to the bank, meaning another institution deciding whether to place money with it for a week, is reading for something narrower again: whether this institution can pay next month without needing that lender to roll over. And a depositor, in the end, is reading nothing at all. Almost no depositor has heard of the measure. Depositors are the reason the whole apparatus exists. A buffer requirement is a rule written for people who will never read it, enforced on their behalf by somebody who will.

Where does the standard come from, and what actually binds an Indian bank?

Two separate acts, and running them together is the most common error on this subject. The liquidity coverage ratio is a Basel standard, published by the Bank for International Settlements at bis.org, and that is where the shape of the measure comes from: a stock of monetisable assets over a modelled thirty day stressed net cash outflow, built from categories, factors and a cap on what may be counted as arriving. Naming that origin tells a reader where the idea was born. The origin tells an Indian reader almost nothing about what any bank in India actually has to do.

The binding rules for a bank in India come from the Reserve Bank of India at rbi.org.in: which assets may be counted as liquid, at what haircut, with what run-off factor by category, with what inflow factor and under what cap on inflows, on what reporting cycle, and from what date. The six factors used in this guide correspond directly to those items. Every single one of them is set in the Indian rule. The global standard supplies the frame and the Indian implementation supplies every number the answer is actually made of, so an account naming only the first has handed the reader an empty box.