Comments on By , 2009)

advertisement
Comments on
The Rationale for and Limits of Bank Supervision
By
Thomas F. Huertas
(Prepared for an FMG Conference at LSE on January 19th, 2009)
This paper is clear, thorough, and sensible, and it also contains some policy recommendations the
radicalism of which is masked by the moderate tone in which they are set out. It is a typical Tom
Huertas paper, and a pleasure to discuss.
These remarks are in three parts. First and briefly I highlight two points Tom makes which are
extremely important and merit further support. Then I turn to some modest disagreements. Finally I
seek to expose the full radicalism of his policy recommendations, and consider if they still seem
sensible and desirable when thus exposed.
But before that there is a question that runs through the paper but is never addressed. Why are we
concerned with bank supervision and regulation if we can close banks while they are still solvent,
protect depositors, and close off outstanding contracts without disorder? I simply note that at this
point, and return to it later.
Key Areas of Agreement
Tom points out that it is important to know “what the authorities would do if an institution required
resolution”. In other words, it is helpful if policy is predictable. The example of the change of
approach between the closure of Bear Sterns and that of Lehman’s demonstrates the importance of
that. Because there was a general and reasonable expectation that the second would be treated like
the first, no-one prepared for any other possibility, and of course substantial problems ensued. The
different treatment of the securities that had been issued by Northern Rock and Bradford and
Bingley, when they were building societies, as permanent interest bearing shares, is a UK example.
Should a building society wish to raise such an extremely useful form of capital in the future, it has
been made more difficult for it by that inconsistency.
This is of course an example in a different context from usual of the benefits of pursuing time
consistent policies. Governments have at least learned to talk as if they understand and accept the
argument in a macroeconomic context; the benefits of it in a microeconomic context are also
manifestly worth pressing. After all, as Tom points out, such policies affect the “liquidity and capital
of banks”. This very important argument certainly merits a paper by itself to develop it more fully.
The second point I would wish to reinforce is the observation that banks are “international in life but
national in death”. This is something on which David Mayes and I have written several times. (See
for example Mayes and Wood in Siklos (2009).) Whatever agreement there may be on information
exchange, however frequently colleges of supervisors meet in a collegiate fashion, all is pointless
unless there is advance agreement and announcement of whether the government of the country in
1
which an international bank is based will or will not take responsibility for the liabilities of its banks
should one of them get into difficulties. Had that been the custom, people would surely have looked
a little more thoroughly at Icelandic banks, and perhaps at some other countries’ banks also. It is
manifestly difficult to bind governments to future actions; but raising the issue would be helpful, and
some promises are more plausibly fulfillable than others.
Some Disagreements
My first disagreement arises at the very beginning of Tom’s paper. He writes that “If society wishes
to regulate banks it must supervise banks”. That is surely not correct. In Britain many firms are
regulated. Take for example Marks and Spencer. That large retailer has many aspects of its conduct
regulated. How it treats its shareholders is regulated. The conduct of its board is regulated. The food
it sells is regulated. But we do not supervise it in the way we supervise banks. The reason is that
when a bank gets into difficulties, perhaps fails, there are very substantial negative externalities. If
such externalities are present at all in the case of another kind of firm, they are comparatively
modest, and have the offset of the benefits that firm’s competitors gain from its failing.
Now, when we spot an externality, even one big enough to be worth doing something about, the
most desirable response is not regulation but to internalise the externality, to get it priced or
handled in some other way. One of Tom’s radical recommendations is actually a big step to towards
doing that, so I return to this point later.
What of supervising to ensure adequate liquidity and capital? Here the behaviour of the central bank
is important, and neglected in Tom’s discussion. If a central bank is standing by to act as lender of
last resort, then, as Thornton, Bagehot, and Hawtrey have all pointed out, there is no need to
supervise for liquidity or solvency. If a bank needs liquidity and can not get it in the market, then it
goes to the central bank and if it can offer acceptable collateral (and the nature of what is
acceptable changes in a time of general liquidity shortage) it gets liquidity. The taking of collateral
expressly removes the need to evaluate solvency. All this of course holds so long as an individual
bank can be closed in an orderly fashion. This can be done in the USA, and will soon be able to be
done in the UK. So subject to that caveat, which is considered below, there is no need to supervise
for liquidity.
That does of course presume that the central bank knows what to do. It did in the UK for many
years; with regard to Northern Rock it did not live up to its past good behaviour, and lent, to quote
Bagehot on its conduct in the Overend, Gurney Crisis, “hesitantly, reluctantly, and with misgiving”.
But it now appears to have relearned its own history. The central bank and its behaviour are of
course absolutely crucial in this matter, because only a cloakroom bank can be sure of having
enough liquidity without central bank support.
Turning now to the supervision of capital, Tom very elegantly argues that this is akin to the
imposition of covenants by a private sector lender, and is justified by the implicit taxpayer support
banks receive. Once again arises the question of why we need provide implicit support if prompt and
2
orderly bank closure is possible. Suppose for the moment, though, that such closure is not possible,
and that we therefore have to supervise. In his discussion of the treatment by the FSA of Northern
Rock, Tom suggests that the problem was not enough supervision. That is a little hard to accept
when one notes that in the year Northern Rock failed the FSA examined Northern Rock’s risk
management procedures and its capital position, and allowed it to pay a special dividend out of
capital. And that at a time when the FSA had been expressing concern about dangers in financial
markets and the underpricing of risk.
The problem surely was inadequate understanding of risk models and of the data on which they
were based. Both bankers and supervisors were guilty of this, failing to understand the endogenous
nature of risk, the importance of fat tails in risk distributions, and the difference between data and
information. The first two have been discussed extensively at this conference and elsewhere. On the
last I would simply point out that a large number of data points drawn from a short and
homogeneous period may give a good statistical fit, but they convey less information than a smaller
number drawn from a longer and less homogeneous period. A hundred annual observations may
give more information than a thousand daily ones.
The particular relevance of this in the present context is that many of the risk models that were in
use were estimated over the past ten or so years. That was a period of low inflation, steady growth,
and low and stable interest rates – the years of what was called in the United States the “Great
Moderation”. These years were hardly representative of a normal state of the world, and their use in
statistical work led both bankers and regulators into complacency. The moral of this is of course that
regulators should do better; but then everyone should do better. What we want is to see if changes
can be devised which make the banking system more robust to the inevitable failures by individuals
and institutions.
This leads me to the final section of my comments.
Concealed Radicalism
Tom’s radicalism comes in the final two sections of his paper. In “containment to cure” he observes
how hybrid capital, which is treated as debt by tax authorities and as equity by depositors, helps
keep down the costs of capital for banks. Tom observes that the use of this is limited by the amount
of tier 1 equity, and proposes that it should therefore be replaced by contingent capital, which
would be converted to equity at the option of the bank’s supervisor.
This is closely related to the situation that prevailed in the 19th century, when bank shares carried
double or triple liability. Bank shareholders were called on to provide additional funds, up to the
original or twice the par value of the shares they held, in the event of insolvency of the bank. This
could be brought back, with the difference that the funds could be called on, under pre-announced
conditions, by the supervisor. Doing so would surely increase caution and thus stability, as well as
providing more capital when needed.
3
It could also have an additional advantage. It would tend to limit the size of banks. It may be
objected that big banks can be more diversified than small ones, and so are more robust to shocks.
This misses the point that when they do fail they cause problems more than proportionate to those
which would follow the failure of a small bank. In addition, the diversification argument needs to be
treated with caution. While banks get bigger and diversify, but remain different from each other, it is
correct. But when they get bigger and diversify so that they all end up looking the same, they are all
exposed to the same range of common shocks. In that case there is a clear disadvantage to having
big banks.
It has been remarked at more than one point in these comments that matters would be much
simpler if banks were capable of being closed in a prompt and orderly fashion. We shall soon be able
to do that in the UK. When we can, the only reason for supervision would be to ensure that banks
were not being too optimistic in their calculations of how much capital they had as they approached
the capital ratio where they would be taken over by the regulators in a “prompt corrective action”.
But although there is no mention of this in the UK’s legislation as at present being considered by
Parliament, that still leaves the problem of big banks. No doubt a bank the size of Northern Rock
could have been handled by these “prompt correction” procedures, but in trying to deal thus with a
large international bank there would be very great difficulties, difficulties which might indeed prove
insuperable in the very limited time available for such actions. So if banks became smaller, bank
failure would be less troublesome and bank supervision and regulation simpler and easier. Tom’s
capital proposals would help move us towards this desirable situation.
There is further radicalism in the section headed “The Limits of Regulation and Supervision”. Here
Tom observes that central bankers frequently comment on the conduct of supervisors. Why, he asks,
should supervisors not comment on the behaviour of central bankers? They certainly can have
reason to do so.
Inflation is bad for financial stability; this is a well-known point. It has been made often, by Anna
Schwartz for example. But there can be other grounds for comment or complaint. A supervisor may
observe inflation at a satisfactorily low and stable rate, but be concerned about the behaviour of
asset prices. They may be rising rapidly, perhaps there is a bubble, and the supervisor fears that
when the bubble bursts there will be financial stability problems.
From our present standpoint that is a most obvious concern. But dealing with it is not easy. Do we
know a bubble when we see it? Suspicion is not the same thing as knowledge; and there is still
dispute over whether there was a bubble in the US stock market in the late 1920s, or just in a few
particular sectors of it. Second, what tool would the central bank use to deal with the market? If it is
an irrational bubble, which policy instrument can deal with it? And if it is the result of a genuine
(albeit possibly mistaken) change in the assessment of future prospects then there has been a
change in the rate at which the future is discounted, a real variable, and therefore capable of being
affected only temporarily, if at all, by monetary policy. (See Wood, 2000). And third, what if the
central bank in question has an inflation mandate, and is complying with it? Unless it possesses
some other policy instrument that it can adjust while leaving its inflation-controlling instrument
unchanged, it is powerless to do anything about the concern even should it share it.
4
Concluding Remarks
Tom’s is a fascinating and thought – provoking paper. Both of its two radical recommendations are
highly desirable. It is indeed a good idea to go back to the 19th century, at least insofar as the
management of bank capitalisation is concerned. This would increase the robustness of banks, and
in time tend to making them smaller. The grounds for concern over the interaction of asset prices
and financial stability are manifest, and it is good to have the concern raised in this thoughtful paper.
But I suspect that it will not be so easy to deal with this problem as it will be to go back to the 19th
century.
Geoffrey Wood
Professor of Economics,
Cass Business School
References
Mayes, David, and Wood, Geoffrey (2009): “National Central banks in an International System” in
The Future of Central Banking ed. Pierre Siklos. Cambridge University Press, Cambridge
Wood, Geoffrey (2000) “The Lender of Last Resort Reconsidered”, Journal of Financial Services
ResearchVol. 18, nos. 2/3, December. Pp 203 - 228
5
Download