Almost exactly eighty years ago, at the end of 1929, the Macmillan Committee was established. Its members were appointed by Philip Snowden, Chancellor of the Exchequer in Ramsay Macdonald’s government, in the wake of the 1929 crash. Its purpose was to analyse the links between the banks, the financial markets and the wider economy and to recommend how monetary policy should be conducted to prevent the sort of deep recession the economy was then descending into. Its dominant member was John Maynard Keynes – who was largely responsible for drafting the report which appeared in 1931 and which contained many of the arguments that would later appear in his General Theory.

In describing the reasons why the Committee was established here is what the Report – and that effectively means Keynes – said: “Not only was the adequacy of the volume of credit created called in question, but the efficacy of the means of distribution was also the subject of adverse discussion. It may, indeed, be said that at no time since the termination of the historic disputes which followed the Napoleonic wars and which led to the passing of the Bank Act of 1844 has the monetary organisation of our country been the subject of so much criticism as in recent times.”

Distribution of credit
What was true in 1929 is true eighty years later – the efficiency of the means of the distribution of credit is once again called into question and has been, as the Macmillan Committee put it, “the subject of adverse discussion”. This is inevitable given the scale of the problems in the banking sector we have seen over the past two years – both in the UK and across the world. The way in which banks operate will now be different – in the near term because of the damage done to their balance sheets and the continuing fragility of their funding; and in the longer term because of the need to ensure that the chances of another banking collapse like the one we have just seen are much reduced. Today I want to consider how this might play out and what the macroeconomic implications – for monetary policy in particular – are. I think it is likely – and quite probably desirable – that banks will become less significant intermediaries in channelling savings from households to companies and to other households. This poses short term and longer term issues which have (very different) implications for monetary policy.

Banking sector will become smaller
There are several reasons why banks are likely to become less important as financial intermediaries channelling saving from one part of the private sector to another. First, in recent years banks have not held enough capital given the type and magnitude of assets they have acquired. For the UK banking sector capital, relative to assets, in the years leading up to the onset of the crisis in 2007 was around half the level that was typical 50 years earlier and probably around one third the level that was usual 100 years ago. In the US the decline in bank capital over the past 100 years is quite probably even larger. The capital ratio shown here is not weighted by any estimate of the relative riskiness of assets, so it would over-state the decline in what one might call effective capital if assets had become safer. But it is not at all clear that the riskiness of bank assets has become lower. For most of the twentieth century a very much higher proportion of bank assets were held as claims on the government or central bank than has been the case in recent years. And for much of the twentieth century most lending to the private sector was short term bank lending to companies. So bank assets are quite likely to have been both less risky and more liquid in the past than they became in recent decades.

Banks will in future need to hold more capital relative to their assets – that seems inevitable given the consensus amongst regulators and I think it is highly desirable. What impact that will have on the cost of intermediating funds through banks is not easy to judge. The Modigliani Miller theorem would suggest the impact on the overall (weighted average) cost of funds to banks would be zero; there would be no change in the weighted average cost of debt and equity. But there are many reasons (largely to do with asymmetric information, deposit insurance and tax) why that is not likely to hold. So it is very likely that the cost of bank funds will rise and as a result less saving will be channelled through them. The Modigliani Miller theorem says that in a market with no distortions changes in the composition of funding between debt and equity do not change the overall cost of finance for a company.

A second reason why banks will probably become less important in intermediating funds within the private sector is because they have not held enough liquid assets. That is assets that are reliably liquid under stressed conditions. Reliably liquid assets are not claims on the private sector. They are claims on the central bank and government debt. Data shows that holdings of these most liquid assets have fallen from about a third of bank assets 50 years ago to a tiny proportion of assets (at least until the recent policy of quantitative easing, which has boosted the reserves of banks at the Bank of England).

These two things – banks coming to have lower capital and less liquid assets – reflect, to some extent, the subsidy generated to intermediation through banks by implicit and explicit insurance given by the state – though they almost certainly also reflect excessive optimism that steady growth and rising asset prices would continue and that because of this banks would have no problems continuing to finance acquisition of assets by rolling over short maturity debt. The belief that there was state insurance of banks – a belief which was probably widespread across the large economies and which would have made it easier for banks to raise funds for a given level of capital and a given degree of liquidity – has been justified by the actions of governments which have supported banks (and on a massive scale). The commitments of the government of the UK to its banks (including those channelled through the Bank of England) are around 60% of GDP. Such commitments are very substantial – though lower – in several other large developed economies. But the UK does stand out.

Maybe it has always been thus. Tacitus describes how the Emperor Tiberius felt compelled to sort out a banking crisis in 33 AD by “distributing throughout the banks a hundred million sesterces, and allowing freedom to borrow without interest for three years, provided the borrower gave security to the State in land to double the amount”. He goes on to say that “Credit was thus restored, and gradually private lenders were found.” If that is accurate then the resources applied to sorting out that bank crisis were relatively small. At the time the average daily wage was probably around 3 sesterces. So the value of the resources used in the support measures was around the equivalent of 35 million working days. In contrast the value of the commitments made by the UK government today could have a value that is the equivalent of around 8 billion working days – more than 200 times as much. Since the populations of the UK today and of the Roman Empire in AD 33 were probably of the same orders of magnitude it seems clear that today’s support measures are far greater than those of Tiberius.

Cheap funds pumped up banks
In the recent past the belief by banks’ creditors that governments would support those banks if serious problems arose allowed banks to access funds at a lower price than would otherwise have been available – given their capital and given the quality of their assets. This subsidy acted to increase the size of the banking sector. And the banking sector in the UK has indeed grown enormously. Banking sector assets – relative to the size of the economy – were fairly stable at about 50% of GDP in the hundred years from 1875 to 1975, but in the last 40 years have increased about 10 fold. They now stand at about 500% of annual GDP. Much of that increase reflects UK banks’ overseas business. But even if we exclude foreign currency assets there has been a huge increase in the size of bank balance sheets relative to the economy; they have risen roughly five-fold since 1980 (see Fig.1).

As banks have grown relative to the size of the economy the composition of their assets (and their liquidity) has changed. One striking aspect of this has been the enormous reduction in the proportion of assets that are government debt. Between 1950 and 1980 on average around two-thirds of banks’ overall supply of credit to the domestic private and public sectors was in the form of holdings of public debt, but that has now fallen to close to zero.

So a range of factors account for the enormous growth of banks: banks have been allowed to hold less capital than they used to and many seem to have actively aimed to minimise their capital requirements under Basle rules; they have come to hold less liquid assets than they used to; a belief that banks would be supported by the state has been confirmed by past events and has been vindicated by how the current crisis has played out; the tax system has favoured debt financing; and the recent period of steady growth and rising asset prices fostered confidence that bank balance sheets were fundamentally strong.

Many of these factors will change and some of them have changed already. The optimism about the fundamental strength of bank balance sheets has evaporated; capital and liquidity requirements are being increased; and it seems natural to now question why debt finance should be subsidised, particularly if it not only costs tax revenue but also increases risks of the government using tax payers’ money to deal with the impacts of higher leverage in the banking sector.

Impacts of smaller banking sector
So it is likely that banks will become less important. What will be the impact of bank assets being smaller as a proportion of GDP and what should the response of policy – specifically monetary policy – be? In thinking about this it is useful to distinguish between the near term and the longer term. The long run impact of less saving being channelled through banks depends crucially upon how dependent companies and households are on bank debt, what substitutes to bank credit are available and how close those substitutes are. Looking at the balance sheets of households and non financial companies reveals a good deal about this. Bank loans finance a much smaller part of the corporate stock of assets than does equity (about 20% relative to around 60%). But the importance of bank credit has increased from around 10% at the beginning of the decade to over 20% in 2009. That rise in the importance of bank debt went sharply into reverse this year when companies have been repaying bank debt and issuing substantial amounts of new equity and corporate bonds.

Households have fewer options for outside finance than do companies. But they have also relied to a far greater extent upon savings for asset accumulation. (Household savings are ultimately the only source of net asset accumulation, though gross asset accumulation can be financed by debt). A comparable measure for household reliance upon debt in accumulating gross assets should take account of the importance of human capital (the future value of labour income). Corporate balance sheets implicitly value the future stream of corporate earnings through the market value of their assets. We can do something similar for households by estimating the present discounted value of future labour income and adding that to the value of financial and tangible assets to form a comprehensive measure of gross wealth.

The stock of bank loans to households is estimated to finance about 7% of assets, though if human capital were excluded, the figure would be around 20%. How costly might it be for households and non-financial companies to find alternatives to bank debt? We can put the question a different way by asking what might be particularly efficient about channelling savings within the private sector through the intermediation of banks. One answer is that banks specialise in assessing and monitoring the creditworthiness of potential borrowers and that there are substantial economies in scale to that. Douglas Diamond, amongst others, has shown that in principle there are significant gains in having large banks save on the costs of assessing borrowers. By delegating the monitoring function to banks there could be efficiency gains in handling the implications of asymmetric information between ultimate borrowers and lenders.

Asymmetric information
This does not in itself imply that banks should be largely debt financed. But there is an argument that asymmetric information, which is both a cause and consequence of delegating monitoring to banks, may make debt contracts natural – both between borrowers and the banks, and between the banks and the suppliers of finance to them. A fixed debt contract is a more natural one when assets are hard to assess than one which makes returns to an investor depend on asset values.

So this line of thinking says that financial intermediaries shouldplay a role in financing risky borrowing and that we should expect them to be very largely debt financed. Further, this debt may need to be short term – as a discipline device on the managers of banks. If this is right it suggests there is some cost to forcing banks to hold a lot more equity capital – and it is one reason why the simple Modigliani Miller proposition is not likely to hold for banks.

I used to find this set of arguments pretty persuasive. Today I am much less convinced. There may be some natural efficiency in having risky credit intermediated through highly leveraged banks which have short term debt contracts with most of those who supply their funds. But it is very clear that this arrangement also generates costs (stemming from the state needing to provide explicit or implicit insurance) and those costs may not fall upon the banks. Furthermore, to the extent that the retail debt liabilities of banks are insured this undermines the incentives of bank creditors to monitor bank management and, by withdrawing funds, to discipline the management. If providers of wholesale funds are not insured they may provide the discipline – but that may also generate wider costs if they withdraw funds and precipitate a generalised liquidity crisis.

The microeconomic case for having large, highly levered banks making risky loans begins to unravel when system-wide contagion effects become important.

So there is no reason at all to conclude that the potential efficiencies from bank intermediation mean limits on the assets they can hold and on the structure of their liabilities must be inefficient. And it is certainly wrong to think that because households want certainty of value and high liquidity, and want a return that is higher than that on truly safe and liquid assets, then banks must be allowed to try to provide it – something that is only likely to be possible with government subsidies. The argument that banks have to be allowed to hold risky assets while offering safe deposits – or else the returns on that saving will be too low – is sometimes presented as a case against so-called narrow banking. But that argument comes close to saying that tax-payers must provide subsidies to make bank liabilities safer than the balance sheets of banks really allow. This is a bad argument.

There are nonetheless potentially some real costs in having less finance channelled through banks that are largely financed with debt. Those costs are the counterpart to the benefits of devolved, or delegated, monitoring and also reflect the natural tendency of those who provide funding for activities that are not easy to value to prefer a fixed contract (i.e. debt) rather than one that makes them explicitly share in the value of assets they cannot easily assess (i.e. equity).

Holding risky loans
It may be that these efficiency gains from having levered banks intermediate funds are small. I think this is the view implicitly taken by Laurence Kotlikoff, who has argued that banks should not hold risky loans. Instead banks might originate lending but then package the loans into securities which would be sold to unit trusts and mutual funds who would market them to households who would then have claims whose values explicitly varied with the value of the underlying portfolios. There could never be a banking crisis in such a world. This way of looking at things downplays the delegated monitoring efficiencies stressed by Diamond and others. It assumes that what are currently bank loans can become assets with values that can be verified – which probably means traded on secondary markets. There are reasons why this is not easy for many of the assets banks hold – some would go further and say that the raison d’être of banks is to hold assets that cannot be easily traded and to specialise in assessing their worth. Of course that would suggest that the very large quantity of tradable assets (or securities) held by banks do not naturally sit within the banking sector.

Anyway,it is one thing to say that assessing the credit worthiness of a potential borrower is hard and quite another to say that banks have such a comparative advantage, such efficiency, in doing this that to have them do less intermediation is very costly. It is very clear now that the monitoring of asset quality by banks in recent years was done far less well than almost anyone thought likely. And it is not self evident that even when it is hard to assess the value of projects that need funding that a debt contract is always the optimal form of finance. Most of the financing of the investment done in the industrialisation of UK did not come via banks. The construction of the railways (and before them the canals) was not mainly financed by bank loans but was largely through companies issuing their own debt and equity (initially private equity and without limited liability). Nor, for that matter, were dot.com investments of the 1990’s – where there was a huge amount of uncertainty about the value of the projects – largely financed by debt.

My own hunch is that a combination of tax factors and implicit subsidies from state support are very important in accounting for the increasing importance of banks, and that their recent spectacular growth does not primarily reflect the efficiencies created by delegating monitoring to bank credit assessors.

Macroeconomic effects
Because of this, policies which have as a side effect that they reduce the amount of intermediation through banks are not, for that reason, undesirable. But they will have side effects. Their impact might be akin that of a tax on bank intermediation. One effect could be a bigger spread between lending and borrowing rates. To the extent that the squeezed spreads of recent years were a reflection of a too low compensation for risks that people did not understand (and which were under-written by the state) – rather than high efficiency in intermediating savings – then the resulting allocations were not optimal anyway.
The implications of higher spreads between bank borrowing and lending rates might include:

1. less saving through banks – and quite possibly less overall saving
2. less investment financed by banks – and quite probably less aggregate investment
3. less credit to households and potentially a lower owner occupation rate and lower house prices.

Not all of these potential effects are negative: a lower owner occupation rate in no way means a lower standard of living and nor do lower house prices. It is very important to see that none of this means that the growth rate of the economy need be lower. One reason is that negative effects are offset to some (very?) significant extent by the likelihood of less volatility and less serious financial crises. IMF and other estimates suggest that of the order of 10% of GDP may be lost forever as a result of serious banking crises. Suppose they come along every 50 years but that if banks have very much lower leverage, hold more liquid assets and that the banking sector is also smaller those problems only cause a drop in GDP of 5% rather than 10%. Suppose underlying growth in normal times, and with a large and levered banking sector, averages 2.75%. The lower fragility equilibrium with a smaller banking sector is one where we could have growth in normal times of only a little above 2.6% and still get the same long run outcomes. This simple example shows that even if one side effect of a smaller banking sector is that in normal times the economy grew more slowly, that effect could be offset by there being less serious financial crises so that the growth rate over the long run would not be lower, and it would certainly be less variable.

Monetary policy implications
The long run implications for monetary policy of these changes aren’t obvious. It is likely that the link between the central bank policy rate and the effective cost of funds in the economy is affected by policies which make banks safer and the banking sector smaller. Itis plausible that the average effective cost of funds in the economy would be a larger mark-up over the policy rate set by the central bank. That would be likely if banks held more equity capital and held more liquid assets, or if there were to be tax reform that reduced the tax deductibility of interest on debt for companies.

So, for several different reasons we might find the natural short term nominal rate – that is the policy rate set by the central bank – consistent with a particular average rate of inflation was a bit lower. This would not mean that the effective real interest rate in the economy was lower. But it is something the central bank would want to take into account in setting its rate. In the short run, substitutes to bank lending to those who have relied heavily upon it are likely to be limited; and the puncturing of the optimism about growth and asset values that came with the banking crisis may have overshot with the potential to become self-fulfilling.

So policy needs to help the transition to a new world where banks are less important and simultaneously try to prevent the recession worsening and have inflation dip persistently below the target level. Quantitative easing (QE) facilitates this by making it easier for non-financial companies to issue equity and debt – something which naturally happens as investors who have sold gilts to the Bank of England look to replace them with assets that are close substitutes. That does two things: it raises the demand for, and the prices of, corporate equity and corporate bonds; and it encourages issuance.

Quantitative easing
Since QE began corporate bond spreads have fallen sharply; for both investment grade and non-investment grade bonds, to their lowest levels since September 2008 (see Fig.2). And since the beginning of the QE policy the FTSE All Share Index has increased by about 45%. Falling corporate bond spreads and rising stock prices have encouraged companies to raise funds in the capital markets. Cumulative corporate bond and equity issuance in 2009 has been much stronger than on average during 2003-2008. However, net external finance raised by private non-financial companies has been weak. Since the beginning of 2009 UK PNFCs have repaid £45.2 billion of debt with UK monetary financial institutions, while total net capital market issuance has been £38.4 billion.

I believe that QE has helped PNFCs become less reliant on bank debt. While it is larger firms that access the capital markets, this may have made it easier than it otherwise would be for smaller firms to access bank credit. In the absence of QE this adjustment would have been more difficult, and companies may have reduced spending by even more in order to reduce their reliance on banks. This process is part of a necessary adjustment that may well have a good way to go. Because QE helps it, and reduces the risk of undershooting the inflation target, I believe it to be effective. That is why I have voted in recent months for its extension.

The way ahead
We have been through a banking crisis. This has prompted fundamental questioning of the role of banks. That is a healthy response. Right now there is an explosion of analysis, commentary and opinion on all this – from governments, bankers, regulators, academics, journalists. That was true in the aftermath of the 1929 crash that led to the establishment of the Macmillan Committee. I want to end by quoting again from the Report of that Committee. I should say that the members of the Committee formed a diverse group. The two leading members were Keynes and Ernest Bevin, then head of the TGWU and future Foreign Secretary. Two people with more different backgrounds than Bevin and Keynes it is hard to imagine: Keynes, the Etonian whose father was a Cambridge academic; Bevin, the illegitimate son of a labourer who never knew his father, was orphaned at eight, left school at 11 and then worked as a farm labourer. Bevin later recalled being asked as a child to read the newspaper aloud for the benefit of adults in what remained of his family and who were illiterate, something we can be sure John Maynard Keynes was never called upon to do.

Keynes and Bevin became allies on that Committee. But when one reads the report it is the voice of Keynes that comes through very clearly: “Another difficulty besetting our task has been due to the complexity, not to say mystery, in which the problems of finance are involved…. It is no exaggeration to say that to the vast majority even of intelligent people the principles of finance and the theory of money are a closed book. When we turn to the experts for enlightenment we find that they fall into two classes. On the one hand we have those who are engaged in the practical business of banking….On the other we have the theoretical economists who as observers and critics endeavour to expound the rationale of that system. Between these two classes of experts…a certain antagonism is inevitable. The practical man finds in his office that he has to deal with a world ruled not by principles but by compromises and is impatient of the formulas of the study which he generally fails to comprehend. Indeed the epithets “academic”, “theoretical”, “doctrinaire”, commonly applied to the philosophic thinker, have acquired a derogatory innuendo in the parlance of a nation which prides itself on its practical common sense and has an innate distrust of intellectual cleverness”.

Keynes had a great deal of intellectual cleverness and he also had courage and enormous determination, and so did Ernest Bevin. They were both certainly willing to respond to the economic crises they saw by advocating a radically different direction to policy. When thinking about the role of banks as we emerge from this crisis I believe we should do the same. The result of pursuing policies that significantly reduce the chances of another banking crisis like the one we have seen is likely to be a smaller banking sector. That is something that creates transitional problems, but it is not something we should seek to avoid.

This is an edited version of a speech delivered by Professor David Miles, an external member of the Bank of England’s Monetary Policy Committee, at Bloomberg’s London centre on December 16, 2009. For the entire speech see: http://www.bankofengland.co.uk/publications/news/2009/137.htm