Solvency and Liquidity Crises: Two Different Failures

A solvency crisis and a liquidity crisis are two different failures. A bank is insolvent when everything it holds is worth less than everything it owes, so its net worth has gone. A bank is illiquid when it cannot produce cash today although its assets are good. One of them needs capital. The other needs cash, and that cash is already sitting somewhere else.

A bank makes two promises at the same time, and the two are answered by two different questions. The first is whether everything it holds is worth more than everything it owes. The first question is about values, and it carries no deadline at all. The second is whether it can hand cash over this afternoon. The second question is about timing, and it does not ask what anything is worth. A bank can pass either one of those and fail the other, and most of the confusion around bank trouble comes from readers who have quietly assumed that passing one means passing both.

Suvarna Commercial Bank Limited, an invented bank, reports the figures both failures are worked from. The two failures share that balance sheet and exactly one figure standing on it, and not one step of the arithmetic.

What are the two questions a bank is actually being asked?

Put in ordinary words the two questions stay apart, and the moment they are written in the language of a balance sheet they start to sound like the same question. Question one: if everything this bank holds were valued honestly today, would it come to more than everything the bank owes? Question two: if the people it owes money to asked for that money this afternoon, could it hand the money over? The two questions are not two ways of asking the same thing, and the answer to one of them says nothing at all about the answer to the other.

An everyday picture holds the difference in place. A household has a flat worth far more than the loan still running on it. On the morning the school fee is due there is nothing in the account. Nobody looking at that household would say it is poor. The money is simply in the wrong shape today. The flat is worth what it is worth; it just cannot be turned into a school fee by eleven o'clock. A household of that kind, stretched to the size of a bank, is the second failure exactly.

Now run it the other way. A different household has a comfortable balance in the account every single month and a business behind it that has been losing money quietly for three years. The bills get paid on the day they arrive. Nothing visible has gone wrong. The value of the business behind the account has gone, and no amount of paying on time reveals it. Insolvency is the first failure, and it is the one that hides.

What makes a bank insolvent, and what number measures it?

Insolvency is measured against net worth and against nothing else. Suvarna Commercial Bank Limited reports net worth of Rs 24,000 crore and gross advancesThe whole of what a bank has put out as loans, counted before anything at all has been taken off for the loans that may disappoint. of Rs 1,44,000 crore. So losses of Rs 24,000 crore on the advance book would take net worth to exactly nothing, and Rs 24,000 crore divided by gross advances of Rs 1,44,000 crore is 16.67 per cent of gross advances. The single division is the whole measurement. The loss that matters is the one large enough to consume net worth, and the share it represents has to name the advance figure it sits on in the same breath.

The distance matters, so set that loss against what the bank actually reports. Gross non-performing advancesLoans where the agreed payments have stopped arriving, so the lender can no longer treat them as running normally. are Rs 6,480 crore, or 4.50 per cent of gross advances of Rs 1,44,000 crore. Against those, provisionsMoney a lender puts to one side in advance because part of a loan looks unlikely to come back in full. of Rs 4,536 crore are already held, being 70.0 per cent of the gross figure. Rs 6,480 crore less Rs 4,536 crore leaves Rs 1,944 crore of net non-performing advances uncovered. Against net advancesWhat is left of the lending book once the amounts already set aside against it have been taken off. of Rs 1,39,464 crore, the Rs 1,944 crore is 1.39 per cent of net advances. Net worth of Rs 24,000 crore is 12.35 times those Rs 1,944 crore.

Careful readers slip here. Watch the denominator move. The gross ratio of 4.50 per cent is struck on gross advances of Rs 1,44,000 crore. The net ratio of 1.39 per cent is struck on net advances of Rs 1,39,464 crore. The Rs 4,536 crore of provisions has already come out of that base, so it is the smaller of the two. Put both over the same base and one of the two is wrong. Every asset quality figure has to say gross or net, and every share has to name the base it was divided by in the same sentence, or the reader cannot rebuild it.

What makes a bank illiquid when every rupee of its assets is good?

Change nothing about values and ask the second question instead. A loan runs to its own schedule, and no bank can telephone a borrower and ask for the principal by four o'clock. So Suvarna Commercial Bank Limited holds advances of Rs 1,44,000 crore that it cannot call back this afternoon, whatever they turn out to be worth. The bank also holds investments of Rs 60,000 crore, and securities can be sold inside the day or handed over as security for a loan of cash inside the day. So if Rs 9,600 crore of deposits is asked for today, an amount that reads as 5.0 per cent when struck on deposits of Rs 1,92,000 crore and as 4.0 per cent when struck on total assets of Rs 2,40,000 crore, the question in the room is not what the advances are worth. Nothing about the value of anything this bank holds has changed, and it can still fail to pay by the close of business. The second failure is exactly that much and no more.

The everyday version is a caterer who has just finished a large wedding. The money is coming, agreed and undisputed, in ninety days. The vegetable seller wants paying tomorrow morning. Nobody has lost anything and nobody is arguing about what anything is worth. The caterer is simply holding a claim that matures after the bill that has to be met. Banks are that arrangement industrialised: they hold long claims and owe short ones, and the gap between those two lengths is where this failure lives.

Advances of Rs 1,44,000 crore beside investments of Rs 60,000 crore come to Rs 2,04,000 crore against total assets of Rs 2,40,000 crore, so Rs 36,000 crore of the total sits in neither of those two lines. Nothing the bank reports says how quickly that Rs 36,000 crore could be turned into cash, and a residual of that size is exactly the kind of gap that decides whether a demand can be met. The two lines with a known length carry the argument. The residual stays a residual.

Where does the cash go when depositors move it?

The whole distinction turns on where the money goes, so follow it slowly. The Rs 9,600 crore did not evaporate when it left Suvarna Commercial Bank. The money was paid into accounts at other banks, and the settlement balancesEach bank's own account at the central bank. The settlement account, rather than notes in any branch, is where a bank's cash sits, and it is separate from the deposits its customers hold with it. of those banks rose by exactly what this bank's fell. The total held across all of them did not change by a single rupee. Somebody's shortfall is somebody else's surplus, so the cash that would close this bank's problem exists right now and is sitting somewhere else.

A capital hole is nothing like that, and the difference is not one of degree. When a loan goes bad, the money is not sitting in another bank's account waiting to be lent back. It is gone. Nobody has a surplus of it. Somebody has to put new money in, and that somebody has to accept that it may not come back. The asymmetry is the reason one failure can be solved by moving cash between accounts and the other cannot be solved that way at all.

The street market version: a hundred stalls trading through a morning, cash moving from one tin box to another all day. At noon one stall is short and another is flush, and the total cash in the market is exactly what it was at opening. Nothing has been created or destroyed; it has moved. Now compare a stall that has been selling at a loss since April. Nobody received that money, so no other stall is holding it. The market's total is genuinely smaller, and the only fix is fresh money from outside.

Which banks receive the cash does not change the arithmetic. Only the amount that moved has to be known, because conservation is a statement about the total and not about any one account.

Which combinations of the two are actually possible?

Two independent questions produce four answers, not two, and the two that sit off the diagonal are where almost every argument about a bank in difficulty actually happens. Solvent and able to pay is the ordinary state: worth more than it owes, and able to hand cash over today. Solvent and unable to pay is the state this whole subject is built around: every rupee of the book is good and the cash is not there this afternoon. Insolvent and unable to pay is both at once, and it is the state people picture when they hear that a bank has failed.

The fourth state is the one that surprises readers: a bank can be insolvent and flush with cash, settling every claim in full on the day it falls due, right up until somebody reads the book properly. There is nothing paradoxical about it. Paying today is an answer to the timing question. The worth of the loan book is an answer to the value question. The two are measured on different sides of the balance sheet and by different arithmetic. An institution can go on answering the first correctly for a long while after the second answer has already gone against it.

Why do the two failures look identical from outside?

A passer-by sees one thing in both cases: a bank that is not handing people their money. The line of people outside looks the same either way. The notice on the shutter reads the same. The depositor's afternoon is the same. Only the balance sheet separates the two, and a person standing in the queue cannot see the balance sheet. The queue faces a genuine information problem rather than a failure of attention, and moving one's own money is a perfectly reasonable response when the two failures cannot be told apart.

The temptation is to imply that a careful observer could tell. A careful observer usually cannot. The evidence available from outside is behaviour, and both failures produce identical behaviour at the counter. The figures that separate them are published on their own schedule, arrive after the afternoon in question, and are read by very few people even then. Anybody who claims to have diagnosed a bank from the length of a queue has diagnosed nothing.

Solvency vs Liquidity: what do the four rows actually ask?

Adjectives make the pair sound like a matter of degree. Put the two side by side as four questions instead. The question put: whether assets exceed what is owed, against whether cash can be produced right now. Over what horizon: no deadline at all, against this afternoon. The fix: new capital, meaning somebody takes a loss or puts fresh money permanently at risk, against a loan of cash, on which nobody loses anything at all if the assets really are good. Who can supply that fix: shareholders or a buyer for the whole institution in the first case, and any holder of surplus cash in the second, the central bank included.

Read the third and fourth rows together and the conclusion is unavoidable: a loan of cash cannot repair capital, and an injection of capital is a slow and expensive way to solve a problem about timing, so getting the diagnosis wrong means applying a fix that structurally cannot work. This is not a matter of one remedy being better than another. The two remedies act on different rows of the balance sheet. Cash lent against good assets changes the shape of what a bank holds and leaves net worth exactly where it was. New capital changes net worth and takes weeks or months to arrange. Neither is a substitute for the other, and neither is a verdict on the institution that needs it.

How does one failure manufacture the other?

The two are not sealed off from one another, and an account that left them tidy would mislead. A bank that must produce cash today and holds only assets that take time to sell will sell them at whatever price is available today. Any price below the carrying valueThe figure an asset is recorded at in the books, which is the number a sale price gets measured against when the asset finally changes hands. is a real loss, and a real loss lands on net worth. Pressed hard enough, a problem about timing manufactures a problem about value, and haste is the only mechanism that carries a bank across from one failure into the other.

The half everybody forgets is the direction this does not run in. A bank whose net worth has already gone does not become solvent by being handed cash, however much of it and however cheaply. Cash arrives as an asset and as a matching obligation to give it back. Both sides of the balance sheet rise together. The difference between them is the whole of what net worth means, and it sits exactly where it sat before. The arrow runs one way only.

The household version is a fire saleA sale made at whatever price is available today, because the seller needs cash today rather than a good price this month. of a gold chain to meet a hospital bill on a Sunday. The chain is worth what it is worth on Monday morning at a proper counter. Sold at four hours' notice it fetches less, and the difference is not an accounting opinion. The gap is money the household no longer has. Speed itself cost them something, and nothing about the chain changed.

The control below moves one thing only: losses on the advance book beyond the Rs 4,536 crore of provisions already held, so nothing is counted twice. The default setting of Rs 1,944 crore is this bank's own reported net non-performing advances. There net worth is Rs 24,000 crore less Rs 1,944 crore, or Rs 22,056 crore, and what can be pledged today is Rs 60,000 crore. Push the losses to Rs 24,000 crore, being 16.67 per cent of gross advances of Rs 1,44,000 crore, and net worth reaches exactly nothing. Push them to the top of the range at Rs 30,000 crore and net worth is minus Rs 6,000 crore. At every one of those settings the second bar stands at Rs 60,000 crore and does not move. Losses eat net worth and leave what can be pledged today exactly where it was.

What do both failures look like worked from one balance sheet?

The two limbs sit side by side, worked from the reported figures of Suvarna Commercial Bank Limited and using not one number that is not already given above. One figure stands in both columns, advances of Rs 1,44,000 crore, and even that one is put to two different uses. On the left it is the book a loss would eat into. On the right it is the line that cannot be turned into cash today. Beyond that single figure, not one operation crosses over. No division worked on the left is worked again on the right, and the consequence is the one the control further down demonstrates: losses on the advance book move every figure in the value column and leave the Rs 60,000 crore that answers the timing column exactly where it was.

The same Rs 9,600 crore comes to 5.0 per cent once deposits of Rs 1,92,000 crore go under the line, and to 4.0 per cent once total assets of Rs 2,40,000 crore go there instead. A share handed over without its base cannot be rebuilt by whoever receives it. Four different bases are in play: deposits, total assets, gross advances and net advances. Every share above names which one it sits on, and that is not pedantry. Naming the base is the only way to check the arithmetic instead of trusting it.

How does somebody outside the bank use this distinction?

Who sets the bars a bank is measured against?

A bank is measured against five bars, and not one of them holds a value for long. All five are set by the Reserve Bank of India and all five move, so a number copied out today would be an error tomorrow rather than a fact that had merely gone stale. Somebody separating two kinds of failure goes looking for exactly those numbers. Knowing that a bar exists and who moves it is worth more than a number that will have changed, and the sheet below can be filled in from the source in a single sitting.

There is a second half to that, and here it is the more interesting half. The bars exist precisely because the two failures are so hard to tell apart at speed. A requirement to hold cash against a short stressed period is an attempt to make the timing question answerable in advance. A requirement to hold capital against assets is an attempt to make the value question answerable in advance. Why a cell here is blank and why anybody would want the number in it to be adjustable are the same explanation.