Bank Resolution: Bailout, Bail-In and What Sits Between

Bank resolution is what is done with a bank that has failed or is about to, so the deposits and payments it carries keep working. A bailout fills the hole with money from outside the bank. A bail-in fills it by writing down or converting claims already on the bank's own balance sheet. Transfers, amalgamations and private recapitalisations sit between the two.

The arithmetic comes first, because the arithmetic is the part that refuses to move. A bank that has lost more than its net worth cannot pay everybody it owes. The shortfall is a rupee amount fixed by the losses that already happened, and nothing about the remedy touches it. Every tool named in this guide is therefore answering one question with a different name written on the cheque. The tools differ only in who bears the loss and in what order, and both of those have to be settled before anybody needs them. The separation holds the whole way down: the size of the hole is arithmetic, and who fills it is a decision. Mixing the two leads to the belief that one tool is cheap and another dear, when the only thing that ever changed was the address the bill went to.

Whose money sits at the bottom of a bank's liability stack?

Turn a bank the right way up before asking anything else about it. On one side sits everything it holds. On the other sits everybody with a claim on those holdings, and the claims are not all the same kind of thing. Some are promises to named outsiders with amounts and dates attached. One of them is not a promise at all.

The owners of a bank hold whatever is left after every other claim has been met, which is why the first rupee of loss lands on them without any rule saying so. Most readers expect a rule to be doing that work, and there is no rule. Net worth is not a fund set aside. Net worth is a subtraction. The subtraction leaves whatever remains once every claim has been taken off everything held, so when the holdings shrink, the remainder shrinks first and by the whole amount. Nobody has to decide that. A remainder is what the word means.

Try the everyday version before the crore figures. A cousin puts in eight lakh rupees of their own money and borrows twelve lakh from a lender to run a small transport business, and the two together buy vehicles worth twenty lakh. If the vehicles turn out to be worth eighteen lakh, the lender is still owed the whole twelve lakh, due when it was always due. The cousin's eight lakh has become six. Nobody wrote a rule that said so. The lender's claim was a fixed promise and the cousin's was the remainder, and a remainder is the only thing a fall in value has to eat.

Suvarna Commercial Bank Limited, an invented bank, is the same shape with more zeroes. The bank holds Rs 2,40,000 crore of assets at their carrying valueThe amount an asset sits at in an institution's own records, which is not necessarily what it would fetch if somebody had to sell it this week.. Against those it owes deposits of Rs 1,92,000 crore, and that is 80.0 per cent of the assets. Another Rs 24,000 crore is owed as other liabilities and borrowings. And its net worth is Rs 24,000 crore, so the assets stand at 10.0 times the owners' slice.

The middle line got there by arithmetic, and it is the kind of thing that can be checked rather than taken on trust. The Rs 1,92,000 crore of deposits added to the Rs 24,000 crore of net worth comes to Rs 2,16,000 crore. The two sides have to meet at Rs 2,40,000 crore. So the other liabilities and borrowings line has to be Rs 24,000 crore, and it is not a figure anybody chose.

What is resolution trying to protect, if not the bank itself?

Here is the sentence that decides how every tool below reads, and it surprises people. Resolution protects the functions a bank carries, meaning the deposits people can reach and the payments that clearThe passing of a payment instruction between institutions and its turning into an actual movement of money. How and how fast that happens is a separate subject. through it, rather than protecting the institution or the people who hold it. The bank is a container, and a container matters only for what it carries.

The everyday version is a bus route. When the operator running the only route between a town and the district headquarters fails, the town needs the six o'clock bus to run tomorrow morning. The town does not need that particular operator to survive, and nobody standing at the stop with a basket of vegetables cares whose name is painted on the side. Keeping the route and losing the operator costs the town nothing it was actually using. Keeping the operator instead spends public patience on the wrong object.

Once that is accepted, several things that look like failures stop looking like failures. A bank whose deposits are moved somewhere else and whose remaining shell is wound up has not been rescued in any sense its owners would recognise, and it has still been a complete success on the only measure that counts. The depositor's money did not stop working for a single morning, and that is the whole test.

The opposite reading causes real damage, because it makes people evaluate a resolution by whether a name survived. A name is not a function. Deposits people can reach are a function, and payments that clear on time are a function, and those are the two things whose interruption spreads far past the building it started in.

What is a bailout, and whose money fills the hole?

A bailout is the simple one to picture and the hardest one to argue about. New capital arrives from outside the bank. The capital can come from an existing owner willing to put more in, from another institution that sees something worth having, or from public money. The direction is what defines it: rupees cross the boundary into the bank from somewhere else.

Under a bailout the existing claims are left standing, the existing owners are dilutedAn existing owner's share of a business shrinking because new shares have been issued to somebody else, so the same holding is now a smaller fraction of a larger whole. rather than wiped out, and somebody outside the bank is now carrying a loss they did not create. Every depositor is paid as promised. Every other lender is paid as promised. The people who put the new money in have bought a claim on a bank that has just lost a great deal, at a price that reflects the rescue rather than the business.

The objection to a bailout is worth stating plainly rather than politically, because the plain version is more useful. The people supplying the money did not price the risk and did not receive the returns while the bank was earning them. A lender who charged a low rate to a bank did so on a view about that bank. A taxpayer never held a view about that bank at all, and never received a rupee of the interest that was being earned on the risk. When the outcome arrives, the two are treated identically. Whatever ought to be done about that, it is a real asymmetry and it is the reason the other tools exist.

There is a second effect that is easier to miss and matters more over time. If everybody lending to banks comes to expect that outside money arrives when things go wrong, they stop watching the bank. A claim that will be met whatever happens is priced as though nothing can happen, and the price of a claim is the main signal anybody outside a bank ever gets about it.

What is a bail-in, and how does capital appear with no money arriving?

No money arrives. The absent rupee is the whole distinction, and it is the part readers refuse to believe the first time. In a bail-in, claims already sitting on the bank's own balance sheet are written down, or converted into an ownership stake, until the bank has capital again. Nothing crosses the boundary. The bank ends the day holding exactly what it held that morning.

A liability that is reduced is capital created, because capital is what remains after liabilities are taken off assets, so shrinking the liability side and adding to the equity side are one act seen from two angles. Read the arithmetic rather than the label and it stops being mysterious. Assets stay at whatever they are worth. Take Rs 24,000 crore of claims off the liability side and the remainder is Rs 24,000 crore larger. The remainder is the owners' slice. The larger remainder is capital, and it came from a subtraction rather than from a payment.

The everyday version is uncomfortable and that is why it is worth using. Go back to the cousin with the transport business, and suppose the lender agrees to take eight lakh instead of twelve, in exchange for a share of whatever the business becomes. Nothing was paid to anybody. The lender's claim shrank by four lakh, and the cousin's remainder grew by exactly four lakh. The vehicles are the same vehicles. The line between two claims on them is what moved.

Which classes of claim may be treated that way, and in what order, is set by the Reserve Bank of India, and it moves; the block further down names that authority rather than writing the order out. The mechanism does not move: a claim that can be reduced is a claim that can absorb a loss, and a claim that cannot be reduced can only be paid.

What sits between a bailout and a bail-in?

Most of what actually happens to failing banks is neither pole. Three arrangements sit in the middle, and it is worth learning them as a set because they are the ones a reader is most likely to see described and least likely to have a name for.

The first is a transfer. The deposits, and a set of assets to match them, are moved to another institution. Depositors wake up banking somewhere else with the same balance, and the failed shell is left behind holding whatever did not travel. The second is an amalgamation, where the whole bank is folded into another institution, claims and all, and the receiving institution takes on everything rather than a selected slice. The third is a private recapitalisation, where existing owners and new ones put capital in without any outside rescue and without anybody's claim being written down.

All three keep the functions running and none of them makes the shortfall disappear, which is the only thing about them worth memorising. In a transfer, the loss stays behind in the shell with whoever is still standing there. In an amalgamation, the receiving institution has absorbed it and will carry it in its own figures. In a private recapitalisation, the new owners have paid for it in the price they accepted. The arithmetic does not care which of the three is picked.

Does the choice of tool change the size of the hole?

Now put one loss through the whole liability side in rupees and let the arithmetic do the arguing. Suppose the assets of Suvarna Commercial Bank Limited turn out to be worth Rs 24,000 crore less than they are carried at. Against Rs 2,40,000 crore of assets that is a fall of 10.0 per cent, and it is exactly the net worth. A fall of exactly that size lands on a round place in the stack and makes the arithmetic easy to follow. A real fall would land anywhere, and the arithmetic would run in the same way.

Take Rs 24,000 crore off the value of the assets and read what is left. The bank now holds Rs 2,16,000 crore. The bank still owes deposits of Rs 1,92,000 crore and other liabilities of Rs 24,000 crore, and those come to Rs 2,16,000 crore of claims. The owners hold nothing. The bank is solventAble to meet everything owed out of everything held. A separate question from whether the money can be raised in time on any particular day. by a hair and completely uncapitalised at the same time, and those two statements are not in conflict. Solvent means the claims can just be met. Uncapitalised means there is nothing left to absorb the next rupee that goes wrong.

Putting it back on its feet is where the three tools finally separate. Returning the bank to Rs 24,000 crore of net worth costs Rs 24,000 crore, and that sentence is true before any tool has been chosen. Under a bailout, Rs 24,000 crore arrives from outside and the deposits and other liabilities are untouched. Under a bail-in, no money arrives and Rs 24,000 crore of claims already on the balance sheet are written down or converted instead. Under a transfer, the deposits move to another institution with assets to match, and the Rs 24,000 crore shortfall stays behind with whoever is left in the shell.

The number is Rs 24,000 crore in all three. Only the name on it changes. The identical number is the central finding, and it is the one a reader is most likely to arrive without.

The figures worked above are one year of one invented bank. There is no earlier year, no comparable bank standing alongside it, and no failure or cycle anywhere in it. The arithmetic on such a set is exact to the rupee and its reach is almost nothing. One year will not convert into a rate for any other period, it proves nothing about whether one way of running a bank works better than another, and it is not typical of anything. Both of those statements are true at once, and an account that gives only the first is selling something.

Why can the order of absorption not be settled during the failure?

The owners' slice absorbs first and needs no rule. Everything above it is a different question, and the answer to that question has to exist before anybody needs it. The reason is about pricing rather than about fairness: everybody who lends to a bank prices the claim they hold on a belief about where they stand when losses arrive, and an order settled in the middle of a failure is an order written by whoever has the most riding on where it lands.

Look at what a lender is actually doing when they set a rate. The lender is asking how likely they are to be paid, and part of that is a question about queue position. A claim that stands near the front is worth more than one that stands near the back, and the difference is what shows up in the rate. Remove the queue and the rate is being set on nothing. The lender has not become safer; they have become unable to tell.

Now imagine settling the queue on the day. Everybody in the room has a claim, everybody knows exactly what their own position is worth, and the loudest voice is the one with the biggest exposure rather than the one with the best argument. An order arrived at that way is not a rule. The order is a negotiation with a rule's name on it, and it tells every future lender that the position they were sold is not the position they will get.

The everyday version sits in any group that has ever pooled money. Ten shopkeepers in one market put in different amounts for a shared generator on different understandings, and if nobody wrote down who gets paid first when it is sold, the argument at the end is decided by who can hold out longest rather than by what anybody agreed. Writing it down at the start costs nothing and changes everything, and the only moment it is easy to write down is the moment nobody needs it.

An order of absorption exists, and it must be fixed in advance. The content of that order comes from the Reserve Bank of India, and it moves often enough that a copy taken from memory would be wrong rather than merely stale on the morning it was revised. A row filled from the source outlives a row filled from recollection.

Who decides that a bank has reached this point, and on what test?

Not the bank, not the people who hold it, and not its depositors. Most readers arrive with the intuition that a bank chooses when to say it is in trouble, and the flat answer is that it does not. The reason is obvious once said: an institution asked to declare its own failure has every reason to wait, and waiting is the single most expensive thing anybody can do here. Every week of delay is another week of losses accruing against a slice that has already gone.

The test itself, who applies it, and what has to be true before these three tools become available all sit with the Reserve Bank of India. So do the capital minimum the bank was meant to clear, the buffers above that minimum, and the instruments it may have issued that carry a condition allowing them to absorb a loss. Underneath its depositors sits a cover routed to the Deposit Insurance and Credit Guarantee Corporation. Five separate requirements sit outside the arithmetic, and every one of them is read off the authority that sets it, at the moment it decides something.

Leaving those five with their authorities buys arithmetic that outlives the rule book. The Rs 24,000 crore identity in the worked instance is true whatever the capital minimum turns out to be this year, whatever the test says and whatever the cover is set at. An account that wrote those in would have offered five things to unlearn and one thing to keep.

What a single year of one bank cannot show

There is one year and one bank. There is no quarterly series, no split of the advance book by segment or by sector, no maturity buckets, no restructured book, no written-off book, no borrower detail, no branch or employee count, no split of the Rs 36,000 crore of assets that sit outside both advances and investments, and no split of interest expended between deposits and the other Rs 24,000 crore of liabilities. Where one of those would decide an answer, the answer is out of reach, and saying so costs less than inventing a line to fill it. An invented line would look identical to a reported one on the screen.