Commentary 13th Aug 2014

Too small to fail

Professor Stephen King, Department of Economics delivered the following keynote speech at the Centre for International Finance and Regulation's recent Market and Regulatory Performance Symposium.

This is a slight change in pace from the two presentations you have seen so far. In that, what (Adjunct Professor) Rod Maddock (Department of Economics) and I are doing is to really think about government policy and government regulation for the banking system within what I would call an industrial organisation framework. So, the motivating question that we have here really goes back to the financial crisis and why Australia was infected by the financial crisis. And it appears to be, that one of the main mechanisms for infection in Australia, was through the import of wholesale funds.

So, Australia is a net importer of wholesale funds in terms of banking systems around the world. We are fairly reliant on wholesale funds as compared to some other banking systems and what happened in the global financial crisis was that the spread on those wholesale funds diverged significantly. In terms of other industries, the banking system in Australia faced a cost shock. So we want to know what does that does and how should our government be responding to that sort of financial shock or that sort of wholesale funding shock.

Our aim is to build a model that develops wholesale funding for the banking system and then we want to impose on that a range of different government policies and see what comes out. Our modelling approach is slightly different to what I would say is a standard approach to modelling a financial crisis in the banking system. We're not going to be talking about moral hazard or adverse selection, the banks in our system are going to have solid loan books and I think that lines up with the Australian experience.

What we are going to be doing however is allowing for a model that looks at the system both with and without the crisis and the key linkage that we are interested in is "fine, you can bring in certain policies on the banking system to deal with the crisis but what do those policies or expectation of those interventions do to the banking system in the vast majority of time where there isn't a crisis". In particular, how does that affect the rates the depositors get from the banking system, how does it affect the rates that are paid by borrowers to the banking system. So they are the sort of issues that we will be looking at and how does it affect competition.

We have used a basic industrial organisation style model. For those of you who are familiar with the banking literature, there is an older style industrial organisation model in the banking system. The Monti-Klein Model goes back to the early 1970s. Its been extended a bit but in some ways it has been pushed to one side by, let me call it, the 'modern' approaches involving asymmetric information: the sort of things that Joe Stiglitz did his work on and got his Nobel Prize for. We are going to extend the Monti-Klein Model.

We allow for very general forms of imperfect banking competition. We consider a banking system, where we allow banks to set up, banks that can be funded by equity and what I'll call foundation debt (just to separate that out and make that clear, as opposed to all the other forms of debt that banks have) and we are going to allow initially in our base case, for free entry into the banking system. Of course that can be modified. There can be a government policy of licensing for banks, which restricts entry into the banking system. That's one of the policies that we are going to be able to look at in this model.

So, for the base case it's going to be free entry. As I said the loan book is going to be solid so default within our model or insolvency for a bank is going to arise when the cash flow is unable to service its foundation debt and the potential threat is going to arise endogenously in our model when the rate on wholesale funds is high.

Let me just present a couple of quick diagrams, which for those of you with economics degrees will probably bring back some nightmares of first year. Demand and supply with some other funny lines on there. It's worth just understanding a little bit about these because our model is in a sense very simple but amazingly powerful in the sort of results that we get.

So how do we model the domestic banking system in different states of the world or under different cost situations? Well here's a situation, which I will call a normal market. It's a market that exists most of the time. Rho is just going to be the probability of a banking crisis. Given that banks are going to be endogenous in our model, this implies some sort of supply in domestic deposits; there is some cost in turning those domestic deposits into loanable products.

There is a supply of wholesale funds plus there is a marginal cost in turning wholesale funds into loanable products that can be the same or different from the marginal cost of the domestic funds, and there is demand for loans domestically in Australia. And what is going to happen depending on the degree of competition in the banking industry is we are going to end up with an equilibrium where the rate for parties who want to borrow funds is going to be given by our "D".

Those funds are going to be sourced from two different sources. We are going to have domestic deposit funds, which is just a green bit here and in this situation we are going to have wholesale funds given by the blue bit here. There is going to be a difference from the banking sector in terms of the cost of domestic deposits and the cost of wholesale funds. That is going to reflect any monopsony power of the banking system.

So in this model, we are able to have banks having market power on both sides of the market. So they are able to have monopsony power with regards to depositors. In a first best world, deposit and wholesale fund marginal costs would be the same but it is not necessary in our model. It could occur in our model but it is just not necessary and similarly the banks are going to have some monopoly power on the other side. We can change that, we can allow for any degree of competition that you like so if you are Steven Munchenberg from the banking association and you believe that it is perfectly competitive, you can incorporate that in our model or if you are a Fairfax journalist, and have different views of the banking sector, yes we can incorporate your views as well. So it is a very general model in that sense.

Here is the diagram in the case of a crisis. There is a shock to the cost of wholesale funds so really all that's changed between the previous diagram and this diagram is this yellow line, which is a total cost of turning wholesale funds into loanable products, has gone up. We pushed it up high enough so that it is no longer a viable source of funds for the banking system. This is the crisis. In a crisis, the banking system is essentially unable to access at reasonable price the wholesale funds. They can formally access it but they choose not to in equilibrium and so we end up relying on the domestic system. What ends up happening here is that there is going to be a change in the prices that are faced by domestic borrowers and the rates received by domestic depositors.

Now, that by itself you say fine, what's the problem here. The problem here is that when we allow for entry into the banking system and the banking system to design how much equity it wants to have and how much of foundation debt it wants to have, in this situation the banking system will become insolvent. That will be an endogenous choice by the banking system.

So the banks rationally recognise this is a possibility, the investors in banks will rationally recognise that this is a possibility and they will simply build in the possibility of insolvency, in the absence of any government policy.

They will build a cost, the expected cost of insolvency, into their decisions. What does that mean? Well, it means, for example, that there is going to be a premium on debt issued by the banks, the foundation debt issued by the banks will have built in to it a premium reflecting the fact that there is possibility of insolvency.

Similarly, in the rate of return of equity, that is going to be built in. So we can allow for a situation where for example the opportunity cost of debt and equity is identical, we can set it equal to the risk-free rate or quite frankly any other rate you like. That can be identical, but in equilibrium you see a different return on debt and return on equity in the absence of a crisis. So that is all built into the model. So we allow for free entry into the banking system in the absence of regulation and we allow for endogenous capital structure and we analyse government policy.

So what are the results? The interesting thing here isn't the background, that is purely so that you know what game we are playing, to put it in formal economic terms, and what is going to be driving our results.

So first off, what government policies have we got here? Well first off in the absence of government policy, in our base case, there will be a probability of a banking crisis. There will be bank insolvency with a positive probability. In those situations the banks will have a capital structure that does not allow them to pay back all of their foundation debt: with a positive probability they will not have enough cash flow. I should step back a bit, that will arise where that is an optimal solution for the bank. So the banks will choose a cost-minimizing capital structure, even though that may involve the probability or possibility of insolvency. Now what happens if that is unconscionable from the government's perspective? So the government has a problem if the banking system goes insolvent. It has macro spill over effects. So the government is going to want to introduce policies to try to mitigate or reduce the probability of this crisis arising.

So, let's look at what actually occurred in Australia. What actually occurred over a weekend is that the government came out and implicitly guaranteed the banks. It did that in Australia, through the deposit and wholesale guarantees. That's what occurred in a range of other countries overseas. What does that sort of ad-hoc intervention or that ad-hoc guarantee do to the banking system?

So let's now assume that the banking system knows. "OK this is the likelihood, this is likely to occur, if we get into trouble, the government is going to bail us out" quite a reasonable expectations for example in the United States given the savings and loans history that we had there. So what's going to happen? Well the economic logic is pretty simple. The government is going to give the banking system "free money" in a crisis.

Now remember when you are in a crisis you have already lost all your equity. Your bank is insolvent so the equity holders have lost their investment. And debt holders are not getting paid the full value of their debt. What is the government going to do? It is going to step in and it is going to cover the difference between the cash flow and the cost to service foundation debt of the bank. From the bank's perspective that is "free money". What is that going to do? Well, firstly put yourself in the Bank's shoes when you're setting up the capital structure of the bank. You got free money in the situation where you go insolvent and the amount of free money depends on the amount of debt you have.

Hmmm, free money sounds good. I get more free money when I have more debt. That is going to create incentives for me to restructure my banking system in good times and in bad, towards having a higher level of debt and a lower level of equity. How far am I going to do that? Well it depends on whether you believe in the Modigliani and Miller theorem or not. If you believe that there is no cost in restructuring the upfront capital structure of your bank, you will have as much debt and as little equity as possible. If you have other factors such as tax rules that limit the amount or mix of debt and equity that you want to have as a bank, then you are going to trade those off. But you are always going to increase the amount of debt you have to increase the amount of money you get from the government if a crisis occurs.

It is sort of like getting free insurance. The more debt you have, the more free insurance the government is giving. This is going to increase the number of banks, again because you are getting free money in a certain situation, you're getting a gift from the government then that increases the expected profitability of banks. That means that more banks are going to enter. That's good in terms of the depositors; they get a higher interest rate. Good in terms of the borrowers, they pay a lower interest rate, and looks superficially as if we got more competition in the banking system. But that banking competition is being driven by this government action in the situation where there is a crisis and there is, from an economic perspective, a deadweight loss. There is an efficiency cost in here because the extra competition is being driven by an inefficient capital structure of banks. So from a social perspective it is actually a bad thing even though superficially everybody says "wow, look at our competitive banking system, isn't this great?".

So, what else happens in this situation? There are two other factors. Firstly, if we think about the depth of the crisis as being the degree of insolvency of banks, then a bail out policy endogenously increases the depth of crisis. Because banks have the incentive to hold more debt, when they run into cash flow problems, they need to be bailed out more than they would otherwise so level of problem is greater. Secondly, a bail out policy encourages banks to restructure in a way that raises their risks of becoming insolvent. So imagine you have a bank, where its capital structure is such that its risk of insolvency is low. By taking on more debt, as a result of the bail out policy, it is increasing that probability of insolvency. States of the world where a bank previously could meet its foundation debt obligations now become states where it can't meet its debt obligations. So an ad-hoc bail out policy encourages a banking crisis.

Now, before I go on very briefly on the next bit, there are numerous variants of this that we can look at. So for example, what happens if the government said, "Well we are not going to have ad-hoc, but we are going to have explicit bail out and there is going to be a tax on the banks to pay for that explicit, bail out". What's going to happen in that situation? Things change a bit. So let's imagine that the government sets a tax so that it neither makes a profit nor a loss. So its setting the actuarially fair rate of tax on the banks in the good times to pay for the bad times. In that situation you are still going to get the restructuring of debt if the tax does not depend on your actual debt mix that you have as a bank.

Banks are still going to restructure because the tax is fixed but the amount of payout they get is variable. In that situation it's easy to show that you get the reverse entry effect. It will be less competitive, in the sense that you will have fewer banks in the system. Why? Because you're getting inefficient capital structuring. Somebody's got to pay for that. Well the banks are paying for it through the tax, so that creates a barrier to entry. In that situation it looks like the ad hoc bail out except you have the reverse in the normal times. In the normal times you have less competition, borrowers pay higher rates, depositors get lower rates.

The second approach we can take to government intervention, and this is in line with the sort of Basel III type recommendations and some of the recommendations that came out in the interim report yesterday, is a minimum equity ratio. What happens in that situation? Well if you believe in Modigliani Miller, if you believe that a bank can hold 30- 40 per cent equity and it has no effect on their funding cost (and there is certainly a group of people out there who believe that), if you believe that then a minimum equity ratio solves the problem at no cost to anybody. You simply set a minimum equity requirement that makes the probability of crisis low enough that the government can live with it. So if you think banks are going to go insolvent or to have a crisis too often, increase the equity requirement a bit more and that reduces the probability of crisis. Push that equity requirement to the point that you like and that solves the problem and if Modigliani Milller holds, there is no cost.

Now I ran this past Rod, Rod is the former group strategy manager of CBA, and when he got off the floor from rolling around laughing, he told me that he didn't think it was a good assumption. What happens if there is a cost in restructuring the funding needs of banks? Then a minimum equity requirement is at least moving you in the right direction. So unlike an ad hoc intervention policy which encourages crisis, a minimum equity policy or minimum equity ratio will reduce the likelihood of a crisis.

There is still a cost because you have got an inefficient funding mix, so that will mean that there is a barrier to entry for a banking system. There is an extra cost of the banking system. That means you will have less competition in the banking system in normal times. So if your main objective is to solve a crisis, this is a good solution but don't think it's free. The cost of this will be fewer banks in the good times, higher interest rates paid by borrowers and lower interest rates received by domestic depositors when the banking system is not in a crisis.

What other policies can we look at? We can also look at things such as licensing, restriction of access to wholesale funds and so on. So it's a very flexible model. Final point, where are we taking this at the next stage you may notice that the original motivation behind this "too small to fail" was so that we could bring different, smaller financial institutions in to this framework.

We can tell you where that goes although we have not got the formal modelling yet, what it is almost certainly going to do is that smaller financial institutions who exist in the good times and then disappear but are not bailed out in the bad times, small enough so that the government does not intervene to bail them out, they will stabilize the system. So for example mortgage intermediaries, such as RAMS and Wizard home loans were before the crisis, before they were taken over by the banks, they actually have a stabilizing effect on the system.

Why? Because they compete in the good times, that discourages more banks from entering, fewer banks enter, all small players disappear in the bad times and for the banks, because there is less competition in the bad times, there is less stress in the bad times. In a sense the small intermediaries become an automatic stabilizer in the system but that is something we are getting to next. So that is stage two where we include the small intermediaries.

 Questions from the floor:

 "I was just wondering whether you have been able to incorporate in your model things like the impact of Ireland's implicit guarantee and the reason why we had to do what we did in Australia in terms to stop the Australian flight of capital and also around the concept of free money, you said, that did cost the banking sector of course. That was $5 billion that went to government revenue so that wasn't free, so have you done modelling on the impact of what that might have done?"

The driver of government intervention in our model is that the government, for whatever reason, cannot allow simply banks to simply go insolvent and go through some normal bankruptcy procedure so that is just an underlying assumption of our model which the government can then respond to.

In terms of the ad-hoc type of approach, in a sense that it is the policy that we've looked at in the most detail so far. Well you may say its not free money in the sense that there's, the way that the government did it in Australia was to include for a premium or a cost associated with that. What you really care about is what did the banking system believe the government was going to do in a situation of crisis. And it's a lot less clear in Australia than it is elsewhere.

So, did the banking system believe that the banks will simply be allowed to crash? It's clear that it was not the situation for example in the US and a lot of evidence in the US that government did not allow that to occur. If that was the situation in Australia, pre the global financial crisis, so if it was just a shock, the banks woke up, on the Friday when the $20,000 deposit insurance came in and then on the Monday when it then is being ramped up to $1million with the wholesale funding deposit. If that was a shock to the banks, the banks would have said " Wow, we didn't expect that" Financial crisis solved, no distortion to the economy , the government simply stepped in at a bad time and saved the banking system, well done, Kevin you've saved us! We'll do better for the next time. So clearly from the point of view of it being an unexpected government intervention, that works once and we don't think the banks will be so naïve in the future. Now, the banks may have expected that intervention upfront anyway.

In terms of the free money, the banking system is bailed out, yes, there was a cost to the banking system but the net benefit was the banking system. The banking system ended up better off. Now you can say look, the whole financial system, the whole macro-economy ended up better off, it would have had a terrible crisis. That's all fine. But the banking system ended up better off by the government's intervention over that weekend. Now there can be arguments about how much the banking system paid for it. Was it a fair cost? Was it an unfair cost? That does not worry me so much. There was a clear beneficiary. It may be the Australian economy. But it was also the banking system.

I'm not so much worried about the cost either because I am curious about the end results and the assumptions around the behaviour and what they may influence or do not influence so if we look at what measure, what you said that there would be increase competition because there would be more banks or more providers, that wasn't case in Australia in fact it was the reverse. You made a comment about how the banks have taken over some of the mortgage originators, bank west for example, so there was behaviour around that time that was actually contrary to an increasing competition. So my interest and curiosity is just that at the time in the crisis in a way that it evolved, and yes that weekend was important but it was also some weeks before that things are playing out globally, is just what really is the impact on the banking sector in terms of it protecting and managing itself through a crisis.

When I say more competition, think of it as being more competition is in a very long term. When a crisis hit, there will be a lot of activities around, there will be amalgamation of the banks and so on. So I'm not talking about a competition at the point of a crisis.

I'm saying, lets step back and, say crisis may occur once in 50 years, what is the degree of competition over that degree of 50 year period when 49 out of those 50 years, we don't have a banking crisis?

So in that situation will an ad hoc bail out policy will lead to more competition than otherwise in that 50 year period? Not within the crisis itself. I agree because within the crisis that is when the deck of cards is thrown up in the air and you get all sorts of policy to try to reduce and mitigate the effects of the crisis which is exactly what we'll see in any industry that is hit by a cost-shock like this or in the banking system.

I'm wondering if your model extends to the collapse of a single institution or maybe 2 institutions given the concentration of the sector or is that really honing and just on major global financial crisis or domestic financial crisis.

Formally, as we have modelled it at the moment, we allow for correlations in the shocks to the banks so, at the moment, in a sense the banking crisis is general so every bank is affected. We can allow for individual banks to face a risk of a crisis due to a cost shock that is only on that particular bank. That's easy to do: we simply change the degree of correlation between the risks. So each bank has a risk of normal conduct and it has a small risk of a crisis arising and we can make that small risk independent between the banks or the other extreme, where in any period any particular bank could face a cost shock with a small probability and that bank could find itself facing insolvency problems.

So we can allow for that range, and in the current modelling that we have got at the moment there is a perfect correlation so it looks at the systemic shock to the system which affects all banks but there is no reason why we cannot move away from that in fact we sort of got 2 versions of this paper at the moment. In the second version we allowed for exactly what you said.

What happens, you're assuming wholesale funds are either cheap or expensive, what happens when there's a continuum between those 2 points and particularly what happens when wholesale markets can actually assess Australian markets higher quality credit than elsewhere so there is not just one price for the wholesale funds?

Extraordinarily annoying mathematically to extend to allow for continuum for states of the world, that could be done. What you would find was that within the model, you would end up with a critical state where if the wholesale funding rate went above that level you'd have a crisis, when the wholesale funding rate is below that level you wouldn't have a crisis.

So to avoid what is not particularly difficult but just an awful lot of notation and mathematics, we said look you can actually capture everything but just defining from the start, two states of the world: crisis state, non – crisis state. In terms of the effect of the crisis and do we face lower wholesale funding cost from the rest of the world for whatever reason, there's no problem with that, that can occur within the model, that is really going to the depth of the crisis rather than anything else. So nothing within our model says that we are better or worse off from the rest of the world.

With the Basel rules, how do they fit in your model? In other words are the capital requirements that are mandated under Basel III rules, have you done any comparison as to whether they are the right amounts basically?

So the key question that I'm not going to answer and is beyond this project is the empirical question that needs to be answered next; is to say, well, what are the costs of remixing or reweighting the funding mix for banks towards higher level of equity or higher equity ratio.

And the theory is quite simple. If Modigliani Miller holds, there is no cost. If Modigliani Miller doesn't hold, why doesn't it hold and what are the costs? Now, Modigliani Miller does not hold in Australia. We have various tax rules to make sure it will not hold but those tax rules are different to other countries.

So what is the cost in Australia of having certain amounts of equity, certain equity ratios of the banks and its not clear to me that the Basel framework has got somebody sitting there saying, "Ah Lets actually work out the cost, let's work out the benefits and lets do a bit of a trade off here". It seems like the Basel III framework, and I am more than happy for the people who are more closely involved in that sort of area than I am to contradict me, but it seems like the framework is being driven by "crisis are bad, let's try and make sure there isn't a crisis" without actually thinking about what does that policy do when there is not a crisis and that's our concern.

What does the policy do the 99 times out of the 100 and our fear is that the sort of situation we get into with Basel III and potentially if Basel III is not far enough you can go even further and there are suggestions in the interim report yesterday that maybe we should go even further. Well that's fine except where are the numbers to tell us what is this going to do in the normal times?

What is this going to do in terms of the decreased degree of bank competition, in terms of the interest rates faced by the borrowers and the depositors and in terms of reduced amount of loanable funds going through the banking system because that would be the inevitable consequence. That is an empirical question and I don't know the answer. If there is any other regulation of that level of import being put in place in Australia, there will be calls saying "show us why the benefits outweigh the detriments, show us that the Hilmer test is being passed" and I haven't seen that been discussed in terms of the Basel III framework and that worries me.

  • stephen-king-economics-pr

    Professor Stephen King

    Professor of Economics Monash Business School

    Read more
    in