The Financial Safety Net: The Layers That Catch a Failure

A financial safety net is the set of arrangements standing behind an institution, reached in a fixed order: the institution's own capital and liquidity first, then the market it can borrow in, then a central bank lending against collateral, then the arrangement standing behind depositors, then the powers to wind the institution down. Each layer catches something the one before it could not.

Every layer answers a different question, and that is the whole reason there are several of them. Capital answers what absorbs a loss. Liquidity answers what meets a payment today. A market answers where more cash can be found, and at what price. A lender standing behind an institution answers what happens when that market is not there. An arrangement behind depositors answers what a particular group is owed regardless of anything else. Powers to wind an institution down answer what happens when none of the earlier answers was enough.

Different questions, different instruments, and no arithmetic between them. The absence of arithmetic between the layers is the part most readers get wrong within a minute of learning that a safety net has layers at all.

Why is a safety net layers in an order rather than one arrangement?

A safety net sounds like one thing. The first correction is the one that matters most: there are five of them, they are reached in a fixed order, and each exists because the one before it cannot do that particular job. The order is not a filing convenience, it is the design. A layer is defined as much by what sits above it and below it as by what it can do on its own.

Think about how a household handles a bad month. Whatever is already in the house gets used first. Nobody has to be asked for it. Only after that does anybody borrow from a relative, and only after that from a lender, and only after that does anything get sold. Nobody sat down and wrote that order out, but it is a real order, and reversing it would be strange. An institution's arrangements are the same shape, except that somebody did sit down and write the order out, and the writing down is what makes it a net rather than a habit.

There is an awkward part, and it arrives at exactly the point where a net would be shown working. A layer that does its job removes the occasion on which it would have been used, so most layers spend their lives having caught nothing. Each layer is therefore learned from what it is designed to interrupt rather than from an occasion on which it interrupted something. Design is the sturdier of the two: a layer's design is a permanent fact about the arrangement, and an occasion is a fact about one set of conditions that will never repeat in that form.

What does an institution's own capital catch that its own liquidity does not?

Most readers merge capital with liquidity, and the merge stays invisible until it costs something. Capital answers a question about size and liquidity answers a question about timing, and the two are measured on different figures.

Take the size question first. If what an institution holds turns out to be worth less than was thought, the difference has to come out of somewhere. The difference comes first out of what belongs to the owners, the figure called net worthWhat is left for the owners after everything owed to everybody else has been counted. It is the difference between two other figures rather than an amount sitting anywhere.. Once the owners' part is exhausted, the difference starts coming out of what is owed to somebody. How far a given fall in value travels depends on leverageHow large what an institution holds is compared with what belongs to its owners, written as a multiple. Higher multiples make the same fall in value read as a larger share of the owners' part.. Leverage is covered separately.

The timing question is a different question about the same institution. The timing question asks whether the money to meet today's obligations is actually there today, whatever the totals say. An institution can be entirely solventIn a state where what is owned is worth more than what is owed, so the owners' part is positive. Being in that state says nothing about whether today's payments can be made on time. and still not have the cash in the room on a Tuesday morning. Most of what it holds is somebody else's obligation, falling due on somebody else's schedule.

Neither one substitutes for the other, and that sentence is worth carrying through everything that follows. An institution with a great deal of capital and very little that turns into cash quickly has not bought itself general safety. The institution has bought a complete answer to one question and no answer at all to the other. The Reserve Bank of India at rbi.org.in sets what an institution must hold on each of the two halves, and both requirements move.

Why does the market an institution borrows in thin out exactly when it is wanted?

An institution short of cash can usually borrow it from other institutions that have some. Borrowing of that kind happens in the interbank marketThe market in which institutions lend cash to one another for short periods, often overnight, usually to square off a day that ended short., and for ordinary purposes on ordinary days it is not one layer among five, it is simply the answer. Somebody is long cash, somebody is short, a price exists, and the day closes.

The property that makes it a layer rather than a solution is that the people in it withdraw for exactly the same reasons as everybody else. The lenders are reading the same conditions, on the same morning, with the same instinct to hold on to what they have until they can see further. So the layer is thinnest at the moment it is most wanted, and it is thinnest precisely because the conditions that made it wanted are visible to the people inside it too.

The everyday version is short. The neighbour who would lend an umbrella is out in the same rain. Nothing about the neighbour has changed, and there is no unwillingness worth complaining about. The umbrella is simply in use.

What makes a lender of last resort different in kind rather than just larger?

Look back at the first two layers and notice what they are made of. The first is made of the institution's own resources, finite and already counted. The second is made of somebody else's willingness, and willingness is exactly what conditions take away. The party providing the third layer is not competing for the same cash as the institutions asking, so the third layer is the first one whose capacity does not shrink when everybody needs it at once.

A capacity that does not shrink is the whole reason the layer exists, and it is why nothing above it can substitute for it however large the numbers above it happen to be. A layer made of participants who are themselves participants is a layer that gets thin in exactly the conditions that call on it. A layer made of a party that is not one of the participants does not. The idea belongs to Bagehot, Lombard Street, 1873, and lending of this kind is done against collateralSomething handed over as security for a loan. If the loan is not repaid the lender keeps it, which is what lets a lender advance cash against assets it has not had time to examine closely. rather than handed over freely. The Reserve Bank of India at rbi.org.in sets what facilities exist, on what terms, against what collateral and at what price. The doctrine itself is covered separately and worked in full there.

Which layer acts on people rather than on the institution?

The fourth layer is the arrangement standing behind depositors, and what it reaches and what it leaves out is covered earlier in this subject. Its position in the order is the odd part. The fourth layer is the only one that acts on a specific group of people rather than on the institution.

Every layer before it is about the institution's ability to keep paying. The absorbing layer is the institution's. The market layer lends to the institution. The third layer lends to the institution. The fourth layer does something structurally different: it settles what one named set of claim holdersThe people and organisations to whom an institution owes something. Depositors are one group of them; lenders to the institution are another, on different terms. is owed whether or not the institution can pay anything at all. Settling a named group regardless of the institution is why the fourth layer can work while the institution is failing, and why the layers above it cannot. The Deposit Insurance and Credit Guarantee Corporation, at dicgc.org.in, settles what a depositor is covered for, and the amount it settles moves.

What is the fifth layer, and why is it a layer rather than a failure?

The last layer is the set of powers to wind the institution down. The decisions those powers take and the order they take them in are covered earlier in this subject. A wind down is a layer rather than a failure of the layers. A wind down is designed, ordered and provided for in advance in exactly the way the four layers above it are, and it is the arrangement's answer once the earlier layers have not been enough.

Designing the last layer in advance changes how the word failure sounds. Reaching the last layer is the arrangement working through its sequence to the end, and it is not the sequence breaking down. A fire door is not a sign that the building has failed. A fire door is the part of the building designed for the case where everything earlier did not hold, and being reached is exactly what it was built for.

What do the two questions look like on one balance sheet?

The worked instance is deliberately narrow. Suvarna Commercial Bank Limited, an invented lender, has layer one worked in full on its figures because layer one is the only layer a balance sheet carries figures for at all. A facility size, an amount available under one and a scheme figure are all settled by an authority rather than by the balance sheet, so everything past layer one is read from design rather than from arithmetic.

Start with the size question. Suvarna Commercial Bank carries total assets of Rs 2,40,000 crore. Its net worth is Rs 24,000 crore, and set against those assets that comes to 10.0 per cent OF ASSETS. Turned upside down, the same fact reads as assets of 10.0 times equity. Now read a decline of 5.00 per cent in what those holdings turn out to be worth. Five hundredths of Rs 2,40,000 crore comes to Rs 12,000 crore, and placed beside the absorbing part that is 50.0 per cent OF NET WORTH. Those two readings are the entire content of the first half of layer one, and they are arithmetic about leverage rather than a statement that anything has happened.

Now the timing question, on the same balance sheet and on a completely different set of rows. Deposits stand at Rs 1,92,000 crore, or 80.0 per cent OF ASSETS. Within them, current and savings balancesDeposits held in accounts the holder can draw on whenever they choose, rather than for a fixed term. They cost the institution less than term deposits, which is the usual reason their share is watched. make up 42.0 per cent OF DEPOSITS, or Rs 80,640 crore, and every rupee of that may be asked for at any hour. Against those, investments of Rs 60,000 crore can be sold. Advances of Rs 1,44,000 crore are somebody else's obligations on somebody else's schedule and cannot be called in at will.

The net worth of Rs 24,000 crore that answers the size question answers nothing at all about the timing question. The two tables have the institution in common and nothing else. Not a row, not a base, not a denominator. The two halves of layer one are measured on different figures, and no operation turns one into the other. The Reserve Bank of India at rbi.org.in settles what is required of an institution on either half.

One block of this balance sheet is worth naming rather than filling. Deposits and the owners' part between them do not account for the whole of what funds it. A third block of liabilities sits between the two, unnamed and unmeasured on either table above. The contents of that block are a question for the institution's own disclosures.

Why is the order of the layers load-bearing?

Each layer is designed on the assumption that the ones before it have been used up first, and that assumption is doing real work rather than sitting there as a formality. If a layer is reached routinely, the layer before it stops doing its job.

Follow the logic once and it is hard to unsee. The first layer is expensive to hold and the third is available, so an institution that can reach the third easily has less reason to maintain the first. The people in the second layer have less reason to price anything carefully. Whether this particular institution can repay them matters less when something behind it can. The effect on behaviour is covered earlier in this subject, and one narrower thing follows from it: the order is what keeps each layer meaning what it is supposed to mean. Reverse it or shortcut it and the layers do not just get used in a different sequence, they change into different things.

How Financial Safety Nets Can Reduce Run Risk without paying anybody?

The reduction has nothing to do with money moving. A safety net does not reduce the risk of a run by paying anybody. A safety net reduces run risk by removing the reason to ask early.

Work through what a person holding a demandable balanceAn amount an institution owes that the holder may ask for at any time, rather than on a stated date. Whether the holder asks today or next month is entirely the holder's choice. is actually deciding. The holder is not deciding whether the institution is sound. A rule that settles claims in the order they are presented makes being early worth something, so the holder is deciding what everybody else holding a similar balance is about to do. Deciding about other people rather than about the institution is what makes the reasoning self fulfilling. The two outcome structure underneath it belongs to Diamond and Dybvig, Bank Runs, Deposit Insurance and Liquidity, 1983, and is worked in full earlier in this subject.

Now put an announced layer behind that balance. Being early has stopped being worth anything, so the holder who knows the balance will be honoured whatever becomes of the institution is no longer weighing what everybody else will do. And once that weighing stops, the outcome it produced is simply off the table. The layer does its work while sitting completely still. The announcement is the instrument. Nothing has to be drawn, lent, paid or settled, and a net that is never touched has not been untested, it has been working the whole time.

Why can a layer that has never been used not be valued?

Working by sitting still ought to be uncomfortable, and it is worth facing squarely. If a layer works by sitting still, then there is nothing to observe. And an arrangement with nothing to observe cannot be sized from experience.

Having nothing to observe is a genuine limitation, and every claim about how much protection a net provides is a claim about something nobody has measured. Be suspicious of confident numbers in this area, including confident numbers offered in a reassuring direction. The arrangement's success consists precisely in the absence of the event that would have generated the evidence. There is no series to fit, no sample to draw from and no counterfactual to compare against.

The design can be inspected even when the performance cannot, so it does not follow that nothing can be said. The layers themselves, the order they sit in, what each one is capable of, and whether the people it acts on actually know about it are all inspectable today, by anybody, without waiting for an event. The result is an inspection rather than a rating, and an inspection is what is honestly available.

What does a safety net not do, even when every layer holds?

Three things, said bluntly. Without them a safety net reads as an advertisement rather than a mechanism.

A safety net does not make a loss disappear. A loss is an amount by which what was owned turned out to be worth less. No arrangement anywhere changes that figure. Every layer set out here decides who is holding the loss and when they find out, and not one of them reduces it by a rupee. Suvarna Commercial Bank's Rs 12,000 crore illustration is Rs 12,000 crore under every layer in the list.

A safety net does not keep the institution alive. Several of the layers work perfectly well while the institution ceases to exist, and the fourth and fifth are built for exactly that. Confusing the survival of the arrangement with the survival of the institution is one of the more common misreadings, and it is why the fourth layer had to be drawn pointing somewhere else.

And it does not remove the effect on behaviour that its own existence creates. A layer that removes a reason to be careful has removed a reason to be careful, and that is the price of having the net at all rather than an argument against having one. The effect of protection on behaviour is covered earlier in this subject. How much of either anybody should want is a judgement about trade offs rather than a fact about a mechanism.

What has to be true for a layer to actually hold?

Three conditions, and the third is the one readers forget every time.

First, it has to be larger than what arrives. Size is arithmetic, and it is the condition everybody checks. A layer smaller than the demand that reaches it slows the demand down and does not stop it.

Second, it has to be available at the moment it is needed. A layer made of somebody's willingness is not the same kind of object as a layer made of a commitment. The second layer's awkward property is that condition, restated.

Third, and this one decides whether the other two are ever tested, it has to be believed. A layer nobody knows about, or nobody believes in, removes nobody's reason to act early. So it never gets the chance to work by sitting still. Such a layer has to work by actually paying, the expensive way round, and that way requires the whole of condition one. The consequence for anyone examining a protection is that the third question comes first. Belief is the cheapest of the three to check, and it decides whether the other two ever matter.