Maturity Mismatch: The Structural Source of Liquidity Risk

Maturity mismatch is a bank funding long assets with short liabilities, and it is what a bank does for a living rather than an error in its balance sheet. At Vindhya Commercial Bank Limited, invented, the first five ladder buckets are all negative and the one year cumulative negative gap of Rs 9,600 crore is 19.0 per cent of the outflows it sits against. A mismatch becomes dangerous through assumption, not size.

There is a sentence in this subject that sounds wrong on first reading, and everything else depends on accepting it: a bank that shows no shortfall in its near term buckets is not a safe bank, it is a bank that has stopped doing its job. Accept it once and every figure below reads as a description of a working bank. Refuse it and every figure below reads as a fault report.

What is maturity mismatch, and why does every bank have one?

The mechanism is identical in a household and only the scale changes. Start with the household version. A person who has bought a house on a twenty year loan and pays for it out of a salary that arrives monthly has taken on a long commitment funded by a short and repeatable inflow. If the salary stops, the house does not shorten itself to match. The person is not being reckless; that is simply what buying a house on borrowed money means. Now turn the picture round and stand where the lender stands. The lender put out money for twenty years and took in money that can be withdrawn far sooner. The asymmetry between long money out and short money in is not a side effect of banking, it is the entire product.

The formal name for the service is maturity transformationThe service a bank provides by taking money that can leave quickly and lending it in ways that cannot come back quickly., and the formal name for its arithmetic shadow is maturity mismatchFunding assets that run for a long time with liabilities that fall due sooner, which is the normal shape of a bank rather than a fault in one.. Depositors want their money available; borrowers want their money committed. Nobody would take a housing loan that could be recalled next Tuesday, and almost nobody would place savings that could not be touched for seven years. A bank stands between those two preferences and absorbs the difference, and the price of standing there is that on any given morning it has promised more money back than it can collect.

Vindhya Commercial Bank Limited is that intermediary at Rs 96,000 crore of total assets. Its deposit book is Rs 76,800 crore, of which Rs 36,000 crore is current and savings money that can be withdrawn on any morning without notice. Against that it holds Rs 57,600 crore of advances net of provisions, lent for years at a time and not callable on demand.

Notice what the three panels do and do not say. The panels do not say the bank has been careless with its funding. The two promises have different clocks, so a business built on making one promise available and the other promise committed cannot avoid showing a shortfall in the near term. The interesting questions therefore begin after the shape is accepted: how big is it, measured against what, and how much of the measurement is a fact about the balance sheet rather than a choice somebody made.

What shape does the mismatch make at this bank?

The table that writes the shape down is a maturity ladder. Every expected inflow and every expected outflow is placed in a time bucketOne time band of a maturity ladder, defined by its start and end day, into which every flow is placed., the two columns are subtracted, and the difference is the gap for that bucket. Vindhya Commercial Bank Limited runs eight buckets, numbered LB1 to LB8, from one to fourteen days at the near end out to beyond five years at the far end. Building the ladder is set out separately under the maturity ladder. Here it is taken as an object and its shape is read.

Here is the shape. Buckets LB1 to LB5 together cover everything falling due inside one year, and every one of the five is negative: minus Rs 2,400 crore in LB1, minus Rs 1,200 crore in LB2, minus Rs 2,400 crore in LB3, minus Rs 1,200 crore in LB4 and minus Rs 2,400 crore in LB5. Then the sign flips. LB6 is plus Rs 2,400 crore, LB7 is plus Rs 2,400 crore and LB8 is plus Rs 4,800 crore. Five consecutive shortfalls at the short end and three surpluses at the long end is not a pattern the bank chose row by row; it is the same single fact, long assets against short liabilities, seen as a table.

The bottom row of that drawing carries a trap. Look at it before reading on. The cumulative gap deepens through the first five buckets to minus Rs 9,600 crore at one year, then unwinds and finishes at exactly zero in LB8. A reader meeting a ladder for the first time often takes the zero as reassurance. The zero is not reassurance and not even information. Inflows total Rs 96,000 crore and outflows total Rs 96,000 crore because the ladder is a rearrangement of one balance sheet, and one balance sheet balances. A ladder would tie to zero for a bank about to fail on Tuesday, so a ladder that ties to zero at the end says nothing about whether the bank is safe. The information is entirely in the path, never in the destination.

How big is the mismatch, and against what should it be measured?

A rupee figure on its own is unreadable. Rs 9,600 crore of one year cumulative negative gap sounds enormous held against nothing and trivial held against the wrong thing. So the useful move is to divide it by something that means something. Two natural candidates exist, and they give two different and equally true readings.

The first is to read each bucket's gap as a share of that bucket's own outflows. LB1's minus Rs 2,400 crore sits on outflows of Rs 9,600 crore, being 25.0 per cent. LB2 is 25.0 per cent, LB3 is 20.0 per cent, LB4 is 12.5 per cent and LB5 is 16.7 per cent. The second and more useful move is to do the same thing cumulatively: the running gap divided by the running outflows. Through LB1 that is 2,400 over 9,600, being 25.0 per cent. Through LB2, 3,600 over 14,400, again 25.0 per cent. Through LB3, 6,000 over 26,400, being 22.7 per cent. Through LB4, 7,200 over 36,000, being 20.0 per cent. Through LB5, 9,600 over 50,400, being 19.0 per cent. Then 10.0 per cent, 5.9 per cent and finally zero.

The curve is worth sitting with. The line rises to a plateau, holds it, then slides to zero. Everything on the right hand half of the picture is the long assets finally maturing and the outflow denominator swelling underneath a gap that has stopped growing. Show only the right hand half and the bank appears to resolve its own problem over time. Every bank does that on paper and no bank does it in a week.

The second yardstick is blunter and lands harder. Take the one year cumulative negative gap of Rs 9,600 crore and hold it against the equity that stands behind the whole institution: share capital of Rs 1,200 crore plus reserves and surplus of Rs 6,480 crore, being Rs 7,680 crore. The gap is 1.25 times equity. The bank has promised, inside one year, Rs 9,600 crore more than it expects to collect, and that shortfall is a quarter larger than all the capital in the institution, and none of that is an error. The comparison is the cleanest statement of why liquidity is a separate exposure from capital. Capital is not cash, and no amount of capital pays a depositor on a Tuesday morning.

How is a balance repayable on demand slotted into a ladder at all?

Now a question a ladder cannot avoid, and one that most readers never think to ask. Of this bank's Rs 76,800 crore of deposits, Rs 36,000 crore sits in current and savings accounts, being 46.9 per cent of the book. Current and savings balances are non-maturity depositsA balance repayable on demand with no contractual maturity date, being current and savings accounts, which a ladder has to place somewhere.. The contract says the money is available whenever the customer asks for it, so they carry no contractual maturity date at all. A ladder sorts by when cash moves. So where does a balance with no date go?

Take the contract at its word and the whole Rs 36,000 crore belongs in LB1, the one to fourteen day bucket. Do that and LB1 outflows become Rs 43,800 crore against inflows of Rs 7,200 crore, and the bank reports a shortfall of over Rs 36,000 crore in its first fortnight, every fortnight, for ever. The result is not a warning, it is a description of no bank that has ever operated. Current and savings balances do not behave like that. In ordinary conditions the overwhelming majority of that money sits still while individual customers move in and out around a stable core, and the stable core is why the balance carries that name.

So the ladder does something else, and it says so out loud. The ladder uses behavioural slottingPlacing a balance in a time bucket by how it has actually behaved rather than by what its contract says.: the balance is placed by how it has actually behaved rather than by what its contract permits. Vindhya slots 5.0 per cent of the Rs 36,000 crore, being Rs 1,800 crore, into LB1, and spreads the remaining Rs 34,200 crore across LB5 to LB8. The single choice of 5.0 per cent is what makes bucket LB1 minus Rs 2,400 crore rather than minus Rs 600 crore, and it is the difference between a gap of 25.0 per cent of that bucket's outflows and one of 7.7 per cent.

The arithmetic is short and it is the hinge of the argument. LB1's total outflows of Rs 9,600 crore already include the Rs 1,800 crore of behavioural run-off. Removing it leaves the other outflows in the bucket at Rs 7,800 crore. Against inflows of Rs 7,200 crore that is a gap of Rs 600 crore, being 7.7 per cent of Rs 7,800 crore. With the Rs 1,800 crore put back, the outflows are Rs 9,600 crore, the gap is Rs 2,400 crore and the share is 25.0 per cent. One assumption, four times the gap.

Why does the same Rs 36,000 crore sit in four different places?

Here is where a careful reader gets suspicious, and the suspicion is worth taking seriously before it is answered. The same Rs 36,000 crore of non-maturity deposits is treated four different ways across four of this bank's tables. In this maturity ladder it runs off at 5.0 per cent inside fourteen days. In the repricing ladder, set out under interest rate risk in the banking book, the whole balance sits in bucket RB5 at one to three years. In the economic value computation it is given an average behavioural life of 0.5 years. And in the thirty day coverage computation it runs off at a blended 16.2 per cent, arrived at as savings of Rs 26,400 crore at an assumed 7.5 per cent giving Rs 1,980 crore, plus current accounts of Rs 9,600 crore at an assumed 40.0 per cent giving Rs 3,840 crore, being Rs 5,820 crore in total and 16.2 per cent of Rs 36,000 crore.

Four numbers for one balance in one bank in one month. The instinct is to call that a reconciliation failure. It is not. The four tables are not four answers to one question, they are one balance answering four different questions, and every one of the four answers is correct for the question it was asked. When does the cash leave in ordinary conditions? The maturity ladder answers that one. When does the rate on the balance change? The repricing ladder answers that one, and a rate reset and a cash movement are simply different events. What is the balance worth today? A valuation answers it, and a valuation needs an average life. How much of it leaves in a stress? The coverage computation answers it, and a stress is by construction not ordinary conditions.

The last column of that drawing is the sharpest thing in the section. Over roughly the same near horizon, this ladder moves Rs 1,800 crore of the balance and the thirty day coverage computation moves Rs 5,820 crore of it, a difference of Rs 4,020 crore on identical money. Both belong to the same bank in the same month. The ladder describes ordinary conditions and the coverage computation describes a stress, and once that is known, the two figures stop competing. The rule this bank's own record insists on is simple and worth adopting: where two treatments of one balance are shown, the reason they differ comes before the second one and not after it.

What limit sits on the mismatch, and how much room does it have?

A shape that nobody caps is a description. The bank turns this one into something managed by putting a limit on it. The maturity ladder as reported is what this bank calls its structural liquidity statementThe bank's own name for the maturity ladder as reported, which is where limit L9 is measured., and limit L9, one of twelve limits numbered L1 to L12, is measured on exactly one row of it: the negative gap in bucket LB1 must not exceed 28.0 per cent of that bucket's outflows. Every one of those twelve limits is set by the board risk management committee, and every one of them is the bank's own figure rather than anybody's requirement.

The current reading looks calm. The LB1 gap of Rs 2,400 crore on outflows of Rs 9,600 crore is 25.0 per cent, and 25.0 over 28.0 gives utilisationWhat is running against a limit expressed as a percentage of the limit, where anything above 100 per cent is a live breach. of 89.3 per cent. A committee paper carrying 89.3 per cent reads as a limit with about a tenth of itself spare, and nobody stops on a number like that. Now convert it into rupees. The cap is 28.0 per cent of Rs 9,600 crore, being Rs 2,688 crore, against a current gap of Rs 2,400 crore. The headroom on the only liquidity ladder limit in a twelve limit framework is Rs 288 crore, being 4.0 per cent of that one bucket's inflows.

There are exactly two ways to consume it and both are small. Let the inflows in LB1 fall from Rs 7,200 crore to Rs 6,912 crore, a fall of 4.0 per cent, and the gap reaches the cap. Or let the outflows in LB1 rise until the gap is 28.0 per cent of them. Solve outflows less Rs 7,200 crore equal to 0.28 times outflows and outflows come to exactly Rs 10,000 crore, a rise of Rs 400 crore or 4.2 per cent. One fortnight's inflows arriving four per cent light, or one fortnight's outflows arriving four per cent heavy, and a limit reported at 89.3 per cent is a breach.

What happens to that limit when the assumption underneath it moves?

Everything so far has treated the 5.0 per cent behavioural slotting as a fixed part of the furniture. The slotting is not part of the furniture. The 5.0 per cent is a run-off assumptionThe share of a balance assumed to leave inside a bucket, which here is the invented bank's own figure and the output of an unvalidated model., and an assumption is a number somebody produced, not a number somebody measured. So the honest thing to do is to hold every other figure on the balance sheet still and ask what limit L9 does as that one assumption moves.

Call the assumption b. Bucket LB1's outflows are Rs 7,800 crore plus 36,000b, and its gap is Rs 600 crore plus 36,000b. Because b sits in the numerator and in the denominator at once, utilisation moves faster than b does, and that is exactly the effect that makes the reported number misleading. At b of 2.5 per cent the limit runs at 61.6 per cent. At the bank's own 5.0 per cent it runs at 89.3 per cent. At b of 6.11 per cent LB1 outflows are exactly Rs 10,000 crore, the gap is exactly Rs 2,800 crore, the share is exactly 28.0 per cent and utilisation is exactly 100.0 per cent. At 7.5 per cent it is 112.2 per cent and at 10.0 per cent it is 131.6 per cent.

The only liquidity ladder limit in a twelve limit framework is 1.11 percentage points of one behavioural assumption away from being breached, and nothing on the balance sheet has to move at all for it to get there. The 1.11 percentage points reconcile from the other direction too: the outflow arithmetic earlier reached exactly Rs 10,000 crore of LB1 outflows as the breach point, and so does this one. Two routes, one answer.

The same seven points as static text, for a reader who never touches the control

Rupee figures in that table are Rs crore, and every one of them is the invented bank's own working number. Read down the last two columns together and the effect is plain. The assumption moves from 5.0 to 7.5, half as much again, and utilisation moves from 89.3 to 112.2, taking a limit from comfortable to breached.

Who decides the assumption the whole answer depends on?

The instinct here is to look for somebody who failed, and there is nobody. So the governance is worth setting out precisely. Vindhya runs eight committees, numbered G1 to G8. The asset liability management committee, G4, has nine members, meets monthly and is chaired by the chief executive, and it is the body that sets the behavioural assumptions including the LB1 slotting. The board risk management committee, G2, has five members of whom three are independent directors, meets six times a year and is chaired by an independent director, and it sets every one of the twelve limits and accepts or refuses every breach.

Both of those mandates are sensible on their own. Behavioural assumptions are technical and belong with the people who watch the deposit book monthly. A limit states the exposure the board is willing to live with, so limits belong with the board. The gap is not in either mandate, it is in the space between them: one body sets an input and another body sets the cap that input decides the status of, and no paper anywhere in the calendar puts the two facts side by side.

Notice what this does not require. The gap does not require anybody to be negligent, to hide anything, or to be under pressure. The gap survives entirely competent people doing entirely correct work, and that survival is what makes it invisible to a control test: every control worked. The gap is a design property of the committee structure, and the only way anybody finds it is by tracing one number backwards from the limit it decides. Model validationAn independent check that a model is fit for the use it is put to, covering its data, assumptions, implementation and limitations. as a discipline is treated separately in this subject area; what belongs here is only the fact that V1 has never had one and that the assumption route and the limit route never meet.

What makes a mismatch dangerous rather than merely large?

Suppose two banks report the identical shape: a one year cumulative negative gap of 19.0 per cent of the outflows it sits against. One of them is fine and the other is not, and nothing in the number says which. Three things do, and every one of them is a fact about the liabilities rather than about the size of the shortfall.

The first is who funds it. Twenty people can each make a decision in a morning and a million cannot coordinate one, so a shortfall carried by a very large number of small depositors behaves nothing like the same shortfall carried by twenty large ones. Vindhya has a live problem of exactly that kind, and it is measured separately: its top twenty depositors and its share of wholesale funding are each capped by their own limit, and who funds the mismatch is set out under funding concentration. The connection is the only part that belongs here: the same rupees appear in this ladder as timing and in that measure as concentration.

The second is speed. Contractual notice is what stands between an intention to withdraw and cash leaving the building. A term deposit contracted for a year cannot leave this fortnight whatever its holder decides; Rs 36,000 crore of current and savings balances can leave on any morning at all, and the only thing keeping them still is habit. A mismatch funded by money that is contractually stuck is a different animal from an identical mismatch funded by money that is merely in the habit of staying.

The third is the one this guide spends most of its length on: how much confidence anybody has in the assumption that produced the table. A ladder built on an assumption that has been independently examined, challenged and re-estimated is a measurement with a known error. A ladder built on an assumption that nobody has ever checked is a measurement with an unknown one, and an unknown error on a number that decides a limit is not a small thing.

Put the three together and the everyday version writes itself. A household running on one salary has a maturity mismatch: a twenty year commitment funded by an income that could stop with one month's notice. A second household with the same loan, the same instalment and two incomes from different employers has the identical mismatch by size and a completely different one by danger. Nothing on either balance sheet differs. The concentration, the speed and the confidence do.

Who actually reads a maturity ladder, and what do they do with it?

Four different people pick this table up and none of them reads it for the same thing, which is the clearest evidence that the ladder is a working document rather than a compliance artefact.

The head of treasury, Devendra Achar in this invented bank, reads it forwards and only at the near end. He is not interested in LB8 and never will be. He is looking at LB1 and LB2, at what has to be funded in the next fortnight, and at whether the Rs 288 crore of headroom on limit L9 survives the week's known movements. For him the ladder is an operating instrument, and the far buckets are somebody else's problem.

The independent director on the board risk management committee reads it backwards, and reads exactly one row: the utilisation. She sees 89.3 per cent, six times a year, alongside eleven other utilisations. The single most useful question she can ask about that number is not whether it is high but what one input would have to do to move it, and the answer here is 1.11 percentage points of an assumption she has never been shown.

A credit analyst at another institution looking at this bank as a counterparty rather than as an employer reads it for the shape and for the disclosure. Is the behavioural assumption stated at all? A bank that publishes a ladder and says nothing about how it slotted its non-maturity deposits has published a table of arithmetic without the one number the arithmetic turns on. Two banks with identical balance sheets and different behavioural assumptions will report different ladders, so the analyst's honest position is that such a table cannot be compared with anybody else's.

And the household version, one last time. The mechanism does not change with scale. A person with a housing loan and one salary already has the whole of this mechanism. The instalment is the near bucket. The salary is the inflow. The habit of the employer paying on time is the behavioural assumption, and nobody validates it either. The difference between that household and this bank is Rs 96,000 crore and a committee structure, and not one thing in the mechanism.

What does a maturity ladder not show?

Three honest limits, and an account that skips them has oversold its own table. First, the ladder counts contractual flows with a behavioural adjustment and lets every inflow count in full. The ladder does not ask whether an inflow would actually arrive if the bank were in trouble. Asking that is a different computation with different assumptions, and it is set out under the liquidity stress scenario. Neither is a check on the other; they have no mapping between them, and inventing one is how a reconciliation gets published that nobody can reproduce.

Second, the ladder has no thirty day point. LB2 closes at day 28 and LB3 runs from day 29 to three months, so a reader who wants a month has to choose a bucket edge rather than a month. The missing month is a property of how the buckets were cut, and the buckets are the bank's own.

Third, and most important, the ladder is a table of when and never a table of who. The ladder states that Rs 9,600 crore more leaves than arrives inside a year. The ladder does not state whether that money belongs to twenty depositors or to twenty lakh of them, and ownership is what separates a large mismatch from a dangerous one. The ladder is one instrument in a set, and reading it alone is reading one instrument in a set.