Maturity and Liquidity Transformation: The Two Functions Banks Perform

A bank runs two mismatches rather than one. In the first, money arrives for short periods and goes out for long ones, so what is mismatched is time. In the second, what the bank owes can be claimed this minute while what it holds cannot be turned into cash this afternoon, so what is mismatched is saleability. Time and saleability are two separate properties of two different things.

The distinction is doing more work than it looks like it is doing. The two mismatches get called by each other's names constantly, and by people who genuinely know the subject. A common account says that a bank borrows short and lends long, and then says three lines later that this is why it cannot pay everybody at once, as though the second thing followed from the first. The second does not follow from the first. The two are separate facts about separate properties, and a bank can sit a very long way out on one of them while sitting almost squarely on the other.

Pulling the two apart and keeping them apart is the unglamorous work, and one bank's own figures set beside them show how large each mismatch actually runs in rupees. The habit that follows is to take any asset a bank holds and answer two questions about it in isolation, when it falls due and whether anybody will buy it today, with neither answer allowed to colour the other.

What is a bank actually standing between?

Start with the two people. The mismatch is not something a bank invented for its own convenience. The mismatch is the shape of what the two sides want, and both wants are entirely sensible.

On one side is somebody with money they do not need this instant. A household running on one salary, say, with a few months of expenses set aside. The household wants the money to be there on the day the school fee falls due, or the day the scooter needs a new clutch, and which day that will be cannot be known today. The whole point of the money is that it is available, so no interest rate on offer will get it committed for seven years. Offered a choice between a slightly better return and getting the money back on demand, the household chooses getting it back, every time, and is right to.

On the other side is somebody who needs money for a long time and knows it. A shop fitting out new premises. The fit-out costs a lump today and earns it back over years of trading. The money will not have come back within a month, so no honest arrangement can require that shop to repay in one. The shop needs seven years and will pay for seven years.

The two wants cannot be matched directly to each other. If nothing stands between them, the household's money sits idle and the shop's premises never open, and both sides end up worse off than they need to be. A bank stands in between and carries the difference. Carrying that difference is the whole of what a bank does, and it has a cost and a risk of its own.

Are these two words for one thing, or two different mismatches?

Two different mismatches, and separating them is the first thing to do, before either one gets a name attached to it.

The first mismatch is about time. Money comes in for short periods and goes out for long ones. The time mismatch is a statement about dates: about when each side of the balance sheet falls due, and about the fact that the dates on the two sides do not line up.

The second mismatch is about saleability. Money that comes in can be taken back at once, and money that goes out cannot be turned into cash at once. The saleability mismatch is not a statement about dates at all. The question is whether a buyer exists this afternoon, at a price worth taking, for something the bank is holding.

Time and saleability are two axes and not two names for one axis, and the clearest way to see the difference is that any household already applies both of them to its own things without ever confusing them. The same one-salary household has a term depositSavings handed to a bank for a settled length of time, coming back at the close of it instead of on the day the holder decides to ask. What one of these pays, and how that is arrived at, is worked through separately. maturing next month: that is short, and the bank will pay it out on request, so it is also easy to reach. The same household has an inherited share in a plot of land three districts away: that is long, in the sense that nobody expects to touch it for years, and it is also impossible to turn into money by Friday. Finding a buyer for a fractional share of a distant plot takes months. Two different things, and each of them has two different properties, and nobody in that household would dream of collapsing the two properties into one.

A bank's assets have exactly the same two properties, and collapsing them is exactly the mistake that gets made. So hold them apart deliberately from here on. Every time an asset appears below, ask both questions about it and answer them separately.

What exactly is mismatched when money comes in short and goes out long?

The time mismatch is the more familiar of the two. A bank takes money for a day, for a month, or on terms that let the depositor ask for it whenever they like. The bank lends that money out for three years, seven years, twenty years.

Now, the sentence that gets said about this is that the bank borrows short and lends long, and taken literally that sentence describes something impossible: a rupee cannot be handed to somebody for twenty years if it has been promised back tomorrow. The bank is actually standing behind a stream of short arrangements that keeps being renewed underneath one long arrangement that is not renewed at all. The twenty year advance is written once and runs to the end. The funding beneath it is a chain: each link ends, and each link has to be replaced by another one the next morning.

So the bank is not relying on any particular rupee staying put. The bank is relying on the stream continuing to arrive. On an ordinary morning it does arrive. For every depositor taking money out there is another putting money in, and the total barely moves even though almost every individual rupee inside it has changed hands.

Why is it worth doing? The prices. Money lent for a longer period normally costs more than money borrowed for a single day, and the difference between what the bank receives on the long side and pays on the short side is what it earns for standing in between and carrying the risk of the chain breaking. And the risk deserves to be said in the same breath rather than saved for later: if what the bank pays for short money rises while what it receives on the long advance is already fixed, that difference narrows, and it narrows without a single borrower missing a single payment. Nothing has gone wrong with the lending. The funding simply got dearer underneath it.

What exactly is mismatched when claims can be cashed and assets cannot?

The saleability mismatch is the one the rest of the subject leans on. A bank issues claims that can be turned into cash instantly, and it holds assets that cannot.

The saleability question differs sharply from the maturity question. There, the question was when the advance falls due. Here, the question is whether anybody will buy it today at a price worth taking. The two are separate questions with separate answers, and the answer to one says remarkably little about the answer to the other.

Take an advance to a small shop with six months left to run. When is it due? Six months. Can the bank turn it into cash this afternoon? Almost certainly not. To sell it, somebody outside the bank would have to be willing to take on a shop they have never heard of, in a locality they do not know, on the strength of paperwork they have not seen. No such buyer is standing there at three o'clock. The advance is short and unsellable.

Now take a government securityA debt instrument issued by a government. Where such instruments are priced, how their value moves and what a hurried sale does to it are worked through separately. with twenty years left to run. When is it due? Twenty years. Can the bank turn it into cash this afternoon? Very likely yes, in minutes. Everybody in the market knows exactly what it is, there is no shop to investigate and no locality to visit, and one of these instruments is interchangeable with the next. The security is long and it is sellable.

So the thing the bank supplies to its depositors is not only a return on their money, it is the ability to have that money now, and that ability is the product being bought. A depositor is not primarily buying interest. The depositor is buying the certainty that on the unknown day the money is needed, it will be there within minutes, and the bank is the only party carrying the difficulty of making that true while the money is out with the shop.

Can an asset be short and hard to sell, or long and easy to sell?

Both, and this is what settles the argument, because once all four combinations can be filled in, the idea that the two mismatches are one mismatch stops being available at all.

Short and easily sold: cash lent out overnight against a government security. The cash comes back tomorrow and, in the meantime, the security backing it is something anybody would take. Near the top of both axes.

Long and easily sold: a government bond with twenty years to run. Its due date is two decades away and it can be sold in seconds. Long on one axis, comfortable on the other.

Short and hard to sell: a six month advance to a small shop. Due soon, and unsellable this afternoon at any sensible price.

Long and hard to sell: a fifteen year loan against a half built factory. Far away on both axes at once, and that corner is where the two mismatches stack.

The two combinations in the middle are the ones that surprise people. The twenty year bond is easier to turn into cash today than the six month shop advance, and the maturity of an asset therefore says very little about whether it can be sold. Read that ordering again if it feels wrong. The asset due in six months is the harder one to sell. Anybody who treats time to maturity as a proxy for saleability has just got the ranking backwards on this pair, and has lost half the subject in the process.

Why does anybody perform these transformations at all?

Two mismatches listed and then left standing would read like an accusation, and they are not one.

Go back to the household and the shop. Both wants were reasonable. Neither party was being unreasonable or greedy or short sighted. The household genuinely cannot commit money it might need next month, and the shop genuinely cannot repay a fit-out in four weeks. If nothing stands between them, the household's money sits in a tin doing nothing and the shop stays shuttered, and there is no version of that outcome in which anybody is better off.

The transformation is not a trick played on depositors, it is the service they are buying, and they pay for it in the return they accept. A depositor with money in current and savings balancesDeposits sitting in accounts a customer can draw on without giving notice, as against money placed for a fixed period. What such accounts pay and how they are priced is worked through separately. accepts less than they might get by committing the money for seven years. The difference between those two returns is roughly the price of being able to have the money on any morning they choose. The depositor has not been cheated of it; the depositor has bought something with it.

And that reframes the mismatch. The mismatch is not a defect that crept into banking and that better management would remove. The mismatch is the product. Remove it entirely and there is nothing left to sell: the arrangement becomes a locker, charging rent to keep somebody's own notes in a drawer, lending nothing to anybody and building nothing.

Does any of this create a single rupee of cash?

No, not one, and this is the part the rest of the subject leans on hardest.

However many claims payable on demand a banking system issues, the quantity of cash in that system afterwards is exactly what it was before any of them were issued. Both halves of that sentence matter, so take them one at a time.

The bank has created a promise about cash. Promises can be created without limit. Writing one costs nothing but the writing. A bank crediting an account has produced a claim, enforceable, real, and payable on demand.

Not one rupee of the thing being promised has been created. Whatever sits in the system's settlement balancesEvery bank keeps one of these with the central bank, and a payment travelling from one to another lands in it. Where the system's cash sits, and what shifts it about, is worked through separately. and in notes is unchanged by the act of issuing claims against it. Issuing does not manufacture. Issuing only promises.

And here is the consequence the whole subject rests on: that is precisely how a system can be short of cash while every single bank inside it is solvent. Solvency asks whether the assets are worth more than the liabilities, and they can be, comfortably, at every institution in the country. Being short of cash asks whether the thing promised can be handed over today, and the promises exceed the cash by the very nature of the arrangement. On any ordinary day this is fine. The promises are not all called at once, and the cash that does move simply goes from one account to another. The arrangement is not fine on a day when a great many of them are called together, and that day is covered separately.

How much can be asked for, and how much can actually be produced?

Abstractions are cheap, so an actual balance sheet stands against them. Suvarna Commercial Bank Limited, an invented lender, reports the figures that follow at a single date. One reading of a balance sheet is a position and not a trend. A trend needs a second reading, and this record carries only one.

The promise. The bank owes deposits of Rs 1,92,000 crore. Of that, current and savings balances come to Rs 80,640 crore, and those can be asked for without notice. Do the division rather than quoting a ratio: Rs 80,640 crore over Rs 1,92,000 crore of deposits is Rs 42.00/- of every Rs 100.00/- of deposits that can be demanded at any moment of any working day.

Cash that can be produced today. Investments of Rs 60,000 crore can be sold or pledgedLodging a security with a lender to back a borrowing, without selling it, so that it returns once the borrowing is settled. Which securities are eligible for lodging this way, and under what conditions, sits with the Reserve Bank of India. without waiting for anybody. Against the same base: Rs 60,000 crore over Rs 1,92,000 crore of deposits is Rs 31.25/- of every Rs 100.00/-.

Cash that cannot. Advances of Rs 1,44,000 crore come to Rs 75.00/- of every Rs 100.00/- of deposits and cannot be called back this afternoon at any price. Advances are the largest thing on the asset side, and they answer none of a demand for cash today.

So more can be asked for than can be produced, Rs 42.00/- against Rs 31.25/- per Rs 100.00/- of deposits, and the right reading of that is not that this bank is short: it is the definition of the business every bank is in. Sit with the second half of that sentence. A reader who takes the first half and stops makes the mistake set out below.

One absence, named instead of filled. Set the Rs 2,40,000 crore of total assets against the Rs 1,44,000 crore of advances and the Rs 60,000 crore of investments, and Rs 36,000 crore is left over belonging to neither one. No split for that leftover exists in the record, so what sits inside it cannot be established, and neither can how much of it would answer a demand for cash today. The split of that leftover is the single line to ask a bank for, and asking beats a guess dressed up as a finding.

Three divisions and a single subtraction produce every rupee reading, and a pen and one sheet of paper reproduce all four of them. Rs 80,640 crore divided by the Rs 1,92,000 crore of deposits, then Rs 60,000 crore divided by that same base, then Rs 1,44,000 crore divided by it once more. Then, once and nowhere else, Rs 31.25/- taken away from Rs 42.00/-. Nothing beyond those four operations was worked.

Liquidity Transformation vs Maturity Transformation: which one goes wrong how?

Both have been defined on their own first, so now the two can be set against each other properly. Four questions do more work here than two definitions would, so take them in order.

The mismatched property. Time in the first case, saleability in the second. Dates that do not line up, against assets nobody will buy on the day the claims arrive.

How each one goes wrong. The time mismatch goes wrong when the price of short money rises while the long lending is already fixed, so the difference between the two narrows with no borrower defaulting anywhere. The saleability mismatch goes wrong when more claims are presented at once than the assets can be turned into cash to meet, and note that the assets can be entirely sound the whole time this is happening.

The limit on each one. Different measures, and the difference between them is practical proof that the two mismatches are different things. Measures about stable fundingMoney judged unlikely to depart at short notice. Which classes count towards it, and the weight attached to each, belongs to the Reserve Bank of India. held against assets that cannot be sold quickly bound the first. Measures about readily saleable assetsHoldings for which a buyer turns up immediately and the price barely suffers. Which holdings qualify under any official measure, and what comes off their value first, is left to the Reserve Bank of India. held against a short stressed period bound the second. Both are set by the Reserve Bank of India, and both move.

Whether one can exist without the other. Yes, and in both directions, as the four combinations already proved. Match every deposit's maturity to an advance of the same maturity and the time mismatch is settled while the advances remain as unsellable as they ever were. Hold nothing but assets that sell in seconds and the saleability mismatch is settled while the dates on the two sides can still be decades apart.

So the two go wrong for different reasons, are bounded by different measures, and a bank can sit a long way out on one while sitting comfortably on the other. One word for both therefore loses half the subject.

Is fragility the same thing as bad management?

Say the difficult thing plainly. An arrangement in which more can be asked for than can be produced today is fragile by construction, and no amount of good management removes that. Fragility is not a shortcoming that a careful institution avoids and a careless one falls into. Fragility is what banking is.

The wrong inference and the right one sound almost identical and lead in opposite directions, so set the two side by side.

The wrong one: this bank cannot pay everybody at once, so it must be badly run. The right one: no bank anywhere can pay everybody at once, that is what a bank is, and the question actually worth asking is what stands behind the promise on a day when more is asked for than usual.

A test that every institution in the world fails is a test that separates nothing, so the useful question is never whether the gap exists but what sits behind it and how fast it can be brought. And what does sit behind it can be named without a single number attached to any of it: the reserves a bank is obliged to keep ready, what it can pledge, what it can borrow from other banks overnight, and what a central bank can supply through a standing facilityAn arrangement a bank may reach for on its own initiative instead of waiting to be invited into one. Every term, period and condition attached to it belongs to the Reserve Bank of India, and the arrangement itself is worked through separately. or an operation. Each of those is worked through separately, and every parameter of every one of them is set by the Reserve Bank of India and moves.

One more thing belongs here. A depositor who asks for their money is not doing anything improper and is not the cause of anything. A depositor agrees to terms that allow the asking, and asking is the whole point of the terms. A person wanting their own money is not a moral event, and any account of these mismatches that treats it as one has stopped explaining and started blaming.

How does somebody reading a bank's accounts actually use this?

Who sets the measures that bound how far this can go?

Four things named repeatedly above are set elsewhere. An authority fixes each of them and an authority revises each of them, so a figure printed for one would not merely go stale on the morning it moved, it would be plainly wrong. The four appear below as labelled rows carrying the name and the site inside them.

The third row deserves a second look. A rule about assumptions rather than about amounts is the least visible of the four and often the most consequential. The rule governs how a deposit is classified according to how likely it is to leave, and what weight each class then carries. Change an assumption in that row and what every bank has to hold changes with it, without any headline figure moving at all and without anything appearing to have happened.