Resolution: What Has to Be Decided, and in What Order

Resolution is what gets done with a financial institution that has failed or is about to, and it is a sequence of decisions rather than a single choice. The functions that have to keep working tomorrow morning are settled first. Where the losses land is settled second. The fate of the claims still standing is settled third. How fast is settled fourth. The order is fixed long before anybody needs it.

Most readers arrive at this word expecting a menu. Something has gone wrong at an institution, an authority steps in, and the interesting question is taken to be which item off the menu gets picked. The menu picture is the wrong shape, and holding on to it makes the rest of this subject impossible to follow. Four decisions are taken, in a fixed order, and the menu only turns up at the second one. Everything before that second decision is about what cannot be allowed to stop, and everything after it is about tidying up what the second decision left behind.

The four decisions have to be settled before anybody is in the position to need them, so working them over a balance sheet that stands shows them at exactly the moment they are really taken. A resolution being planned is a resolution being planned for an institution that is perfectly fine.

The sequence rests on one fact about what a financial institution is. An institution is two things at once: a business that can fail like any other, and a carrier of functions that other people arrange their week around. An ordinary winding upThe ordinary process for closing a company that cannot pay what it owes, and for getting back what can be got back for the people it owed. How that process runs is settled separately. treats only the first of those, and ranks getting money back to creditors above everything else. Resolution exists because that ranking, applied to an institution, would switch the functions off while the getting back was worked out. Those functions are what the rest of the system is bolted to. Reverse the ranking and everything else in this guide follows.

What has to keep working tomorrow morning?

Before a single rupee of loss is allocated, somebody has to write down which functions cannot be interrupted. The balances people can reach when they need them. The payments that clear through this institution on their way to somewhere else. The obligations that settleThe moment an obligation is actually discharged, when the money and whatever was bought change hands for good. The timetable for that is set separately and is not written here. on a date that somebody else has already built a plan around. The list is short, and writing it is unglamorous work. Most explanations of the subject skip it for that reason.

Continuity has to be first because a function that stops is not recoverable later, and every hour it is off it is reaching the institutions connected to this one. A loss can be argued about on Wednesday and allocated on Friday, and the loss is the same size on both days. A payment that failed to clear on Wednesday is not un-failed on Friday: somebody downstream did not receive money they had already committed, and they have spent Thursday dealing with that. The first decision is first because it is the only one where delay changes the size of the problem.

Here is the everyday version. A small factory changes hands. Whatever the buyer and the seller are arguing about, the gate still has to open on Monday morning. The fifty people who walk through it did not choose the owner, cannot wait for the paperwork, and have their own rent falling due. The gate opening is not a favour to the factory. Opening the gate is the recognition that other people's arrangements are already hanging off it. An institution's functions work the same way, except that the people hanging off them number in the lakhs and mostly do not know they are doing it.

Where do the losses land, and who settles that?

The loss decision is the one everybody thinks the whole process consists of, and it is worth being precise about it. The question is not how much, which is arithmetic and is worked separately. The question is onto whom, taken across the layers of the institution's own funding and moving outwards from the owners' claim. The owners hold the residual, which means their claim is whatever is left after everything owed has been met, so the first rupee of any loss lands on them by the definition of what they hold. Nobody decides that part. The decision proper begins once the owners' claim has gone.

The natural next question is the running order of the layers after that. Which classes of claim absorb loss, and in what order, is set by the Reserve Bank of India at rbi.org.in, and it moves. So the row is drawn with that name inside it and nothing in the value column. The teachable part is why there has to be an order at all. The reason is a fact about people rather than a rule anybody wrote. Anybody putting money into an institution, in any form, is quietly betting on which layer they are sitting in when a loss turns up. If nobody has written the order down, that bet is not a price. The bet is a hope instead. Hopes are not comparable across institutions, and nothing can then be sensibly priced against anything.

What happens to the claims that are still standing?

After the losses have landed there are still claims that nobody extinguished, and every one of them has to end up somewhere. There are exactly three somewheres, and the list is short enough to set out in full, though nobody ever does. A claim is carried on by whatever continues. Or it is moved to somebody else who takes it on. Or it is settled and closed. The three destinations are the whole set, and a claim that has not reached one of them has not been dealt with, it has merely been left.

The part a reader will not have thought of is that this decision, and not the loss decision, is what determines whether the people holding those claims are still customers of anything on Tuesday, and it is taken by somebody who is not one of them. A claim holder may end the week having lost nothing at all in money terms and still find that the arrangement they had is now with an institution they never chose, on terms they did not negotiate. Nothing about that is unfair. The change of counterparty is simply what the third decision does, and it is a good reason to know the decision exists.

Why is how fast a decision rather than a fact about the world?

Speed feels like a constraint. Speed is not a constraint; it is the fourth decision, and somebody takes it. Moving quickly limits how long the functions are exposed and how far the trouble travels, and moving quickly also means answering where the losses land on less information than anybody would like. Both halves of that are real, they pull in opposite directions, and there is no setting that gets both.

The consequence is sharper than it first sounds. A resolution completed over a weekend and one completed over months are not the same instrument run at two speeds. The two know different things at the moment of answering, and the difference shows up in their answers to the second and third decisions. The weekend version allocates loss on estimates. The long version allocates loss on better figures, having left the functions out in the weather while it worked them out. Anybody comparing two such processes and asking which was better run is usually comparing two different answers to the fourth decision without noticing that the fourth decision was taken at all.

Why is the order of the four itself the protection?

The ordering is the heart of the matter, and the reason is about bargaining rather than tidiness. Suppose where the losses land is settled before what has to keep working is settled. Every argument about those losses is now conducted with the functions switched off, or visibly about to be. Whoever is best placed to threaten the functions now holds enormous leverage, and the alternative to using it is absorbing a loss they can avoid by pointing at the switch. So they use it. Settle continuity first and that threat is off the table before anybody sits down at it.

The same mechanism runs in a much smaller setting. Ten shops share one building and one generator. If the electricity bill is disputed before anybody agrees the generator keeps running, the shop that controls the fuel key is going to win the argument about the bill, and everybody knows it before the argument starts. Where the shops agree first that the generator runs, the bill becomes a question about money instead of a question about who can hurt whom. The order in which two questions are answered has changed the answer to one of them without changing a single fact.

One warning, and it covers the easiest confusion on the whole subject: both things are called an order. The order of the decisions is continuity, then losses, then the claims left standing, then speed. The order of the claims is which layer of funding absorbs loss first, second and third, and that one is set by the Reserve Bank of India at rbi.org.in and is covered separately. The two orders are different objects. One is a sequence of questions somebody works through; the other is a ranking of the people who will feel the answer.

Which layer does the second decision actually reach for?

Time to put this on a real balance sheet. Suvarna Commercial Bank Limited, an invented lender in no trouble whatsoever, holds total assets of Rs 2,40,000 crore. AdvancesThe loans a lender has made and still holds on its own books. The kinds of advance and the pricing of each are settled separately. are Rs 1,44,000 crore and investments are Rs 60,000 crore. On the funding side, deposits are Rs 1,92,000 crore at 80.0 per cent of assets and net worth is Rs 24,000 crore at 10.0 per cent of assets, and assets are therefore 10.0 times equity. The ratio of assets to equity is the leverageHow many rupees of assets an institution carries for each rupee the owners put in. Leverage is settled separately and is used here rather than rebuilt. covered separately.

Now look at the funding side as a stack rather than a list. The second decision moves through it in an order. At one end sits the owners' claim of Rs 24,000 crore. The claim is the residual and therefore takes the first rupee of anything. At the other end sit the deposits of Rs 1,92,000 crore. And between those two sits a third layer of liabilities, and that layer is exactly what the second decision reaches for once the owners' claim is gone.

The size of that layer differs from one lender to the next, and no general figure stands in for it. The question is worth asking of any lender: what is sitting between the deposits and the owners, how much of it is there, and on what terms was it taken? The middle layer is where the answer actually gets decided, and almost nobody looks at it.

A control nobody can quote is a control nobody can use. Two readings from it are worth carrying away in words. At 5.00 per cent OF ASSETS the fall is Rs 12,000 crore, or 50.0 per cent OF NET WORTH. At 10.00 per cent OF ASSETS the fall is Rs 24,000 crore, or 100.0 per cent OF NET WORTH, and there the count of layers standing between the reading and the deposits falls from two to one. Notice what the second number is. The fall that exactly exhausts the owners' claim is the owners' share of assets, here 10.0 per cent OF ASSETS, and that is a derived fact about this balance sheet rather than a law about lenders. Rukmini Finance Limited, also invented, holds Rs 18,000 crore of assets against net worth of Rs 3,600 crore, so its net worth is 20.0 per cent OF ASSETS and its assets are 5.0 times equity. The same 5.00 per cent fall reads as Rs 900 crore there, or 25.0 per cent OF ITS NET WORTH rather than 50.0 per cent. Same fall, same arithmetic, different answer, and the only thing that changed was how much of the funding the owners put up.

Can an institution whose books balance still reach this point?

An institution whose books balance can still reach this point, and the possibility is the idea most readers arrive without. Two different tests are applied to the same balance sheet at two completely different speeds. The first asks whether what is owned exceeds what is owed. The first test is answered at a date, carefully, on figures struck at that date, and a comfortable answer feels conclusive. The second asks whether what is due today can be paid today. The second test is answered every single working day, and not by an opinion or a statement: it is answered by whether the money is actually there when somebody asks for it.

An institution can pass the first test and fail the second one. The first is about totals and the second is about timing, and nothing whatever about a comfortable total makes cash appear on a Tuesday. A household can be in exactly the same position. A person can hold land worth many times what they owe and still be unable to pay a bill falling due this week. Land does not turn into money on the day it is needed. Nobody would call that household insolvent. Nobody should call the institution's problem imaginary either.

Which of the two tests actually triggers this process, and who applies it, is set by the Reserve Bank of India at rbi.org.in and is left as an empty row further down. The general comparison between the two kinds of failure is covered separately. The point is only that both tests exist, that they run at different speeds, and that the answer to one of them says nothing about the answer to the other.

Who decides that an institution has reached this point?

The honest answer to the who is short, and it is the part that never moves. Not the institution. Not the people who own it. Not the people who funded it, and not the people who hold balances with it. Everybody in that list has an obvious interest in the answer, and that is exactly why none of them gets to give it. Somebody outside the institution applies a test that was written down before anybody needed it.

The test itself, the powers that follow from it, and every condition attached to each of them belong to an authority. Five labelled rows below name the authority for each and leave the value cell empty. Each row is set by the authority named inside it and can be changed by that authority, so a stated value would be wrong on the day it changed rather than merely old. Take the sheet to the sites named in it and fill the middle column from the source. The answer will then be current, which no printed value can promise.

How is this different from an ordinary company being wound up?

A reader who has met one of these processes will assume the other is the same thing under a different name, and the assumption is close enough to be dangerous. Both processes have the same list of things to settle. The two processes rank that list in opposite orders, and the ranking is the entire difference. An ordinary process puts getting money back to creditors at the top and keeps things running where running happens to help the getting back. Resolution puts continuity at the top and settles the getting back inside whatever continuity allows.

One consequence follows immediately and explains something that otherwise looks like bureaucracy. Because the ranking is reversed, this process needs decisions taken in advance that an ordinary one can take as it goes along. If continuity is the top objective, the work of establishing what has to keep working cannot begin on the morning it stops. Somebody has to have written it down already, while everything was calm and nobody was under pressure. Calm is the only condition under which such a list gets written honestly. Writing it in advance is not paperwork for its own sake; it is the first decision, taken early because it cannot be taken well late.

How does anybody outside the process actually use this?

What has to be true for the trouble to stop here?

The last question is whether the trouble stopped here or travelled on. A resolution that worked is one after which the institutions connected to this one are carrying nothing they were not carrying before. The test has three parts, in order.

The functions did not stop, so nobody depending on them had to do anything at all. The losses landed on people who had already priced for the possibility, so nobody was surprised into acting, and it is being surprised rather than being hurt that moves trouble along a connection. And every claim left standing found one of its three destinations, so nothing was left hanging for somebody else to worry about next week. Fail any one of those and the process has not contained anything: it has moved it, and the arithmetic of what was moved did not shrink on the way.

Which is why the ordering argument set out above is not an academic point about tidiness. Continuity first is what buys the first of the three. A written down loss order is what buys the second. A named destination for every claim is what buys the third. The order of the decisions is not a procedure somebody imposed on the process; it is the only order in which the three things that have to be true can all still be true at the end. Two more parts of the picture sit outside this guide: what a depositor is covered for, and how a landed loss decides which connection gets loaded next. The doctrine of standing behind an institution as a last resortThe idea that a central bank may stand behind an institution rather than let a shortage of money settle the matter. The doctrine belongs to Bagehot, Lombard Street, 1873 and is settled separately., which belongs to Bagehot, Lombard Street, 1873, is settled separately as well.