Memorandum from Julian D A Wiseman
1. The Bank of England is better at implementing
monetary policy than it was, but not as good as it should be.
Unnecessary complexity in the Bank's operations worsened the effect
of the credit crunch on the UK. Better implementation of monetary
policy would be simple and cheap.
2. The Bank of England has traditionally
been abysmal at implementing monetary policy, whether measured
by the variance of short-term interest rates around the policy
or by market participants' opinion of the quality of implementation.
Why does this matter? Under normal circumstances it leads to unnecessary
volatility and a waste of resources, as well as allowing insiders
to manipulate the market to their own advantage. And under crisis
conditions holes in systems tend to widen as participants become
less willing to trust each other's solvency.
3. In 2003, the last full year of the previous
regime, the success of the Bank of England at implementing monetary
policy in £ was worse than the implementation of policy in
dollars, euro, yen, Swiss francs, Canadian dollars, Australian
dollars, New Zealand dollars, Danish krone, Swedish krona, Czech
koruna, and Taiwanese dollars. To be fair, the Old Lady's implementation
was better than that in Norwegian krone, Icelandic krona, and
several, but not all, emerging-market currencies. For a central
bank that thought and thinks of itself as market-savvy, this was
less than impressive.
4. So with this background in mind this
critic of the Bank's implementation of policy was delighted to
hear, on 28 July 2004, the Executive Director Markets quoting
Charles Goodhart's description of an earlier version of Bank operations
as "confused and silly".
Maybe this time the Bank would get it right!
5. In Reform of the Bank of England's
Operations in the Sterling Money Markets, 7 May 2004
the Bank describes its objectives, the first of which says:
The first and primary objective is for sterling
overnight interest rates to be in line with the MPC's repo rate,
leading to an essentially flat money market curve out to the next
MPC decision date, with very limited day-to-day or intra-day volatility
in market interest rates at maturities out to that horizon. Beyond
the next MPC decision date, market interest rates will be free
to reflect market expectations of future MPC interest rate decisions.
6. By "overnight interest rates",
the Bank means the cost of borrowing money against collateral
(collateral being discussed later). So the "first and primary
objective" is for the cost of secured overnight money to
be very close to the policy rate. This is good; and this is measurable.
7. Also of note, because of the margin by
which the Bank failed, is the third objective:
A simple, straightforward and transparent operational
8. In the UK, and in most countries, the
banking sector needs to borrow money from the central bank. Banks
withdraw physical cash from the central bank, andmostlyneed
to pay for this cash with borrowed money. In sterling the banking
system needs to borrow about £28 billion from the Bank of
Englandthe £41 billion banknote issue less the government's
£13 billion overdraft in the Ways and Means account. These
numbers vary along with the size of the note issuelarger
at weekends and over public holidaysbut nonetheless the
banks are almost always in the position of borrowing from the
Bank of England.
9. So why not do this the easy way? To implement
a policy rate of 5% the Bank of England should be willing to lend
money to good counterparties against good collateral, overnight
and in good size, at the policy rate of 5%; and also be willing
to accept overnight deposits from counterparties at a slightly
lower rate, say, 4.9%.
10. That would be enough: against good collateral
the cost of overnight borrowing could not rise above the assumed
policy rate of 5%, and it could not fall below 4.9%. With banks
being net borrowers from the Old Lady, the price would typically
trade near the top of this narrow band. And every commercial bank
would know that collateral is as good as money: if the collateral
is put with the Bank of England, then money can be borrowed at
the policy rate.
11. For reasons of prudence the central
bank's standing facilities should not be in?nite in size. If a
counterparty were to become insolvent the collateral would have
to be sold, and this might entail a loss. Since the banking sector
needs to borrow a total of about £28 billion, a sensible
upper limit for lending to any one large bank might be something
of the order of £20 billion. As the banks are net borrowers,
the limit on a counterparty's remunerated deposit with the central
bank could be smaller: perhaps half the maximum overdraft, so
£10 billion per bank. Larger borrowings would be prohibited;
larger deposits would be unremunerated.
12 This system would have satis?ed the Bank
of England's objective of "simple, straightforward and transparent".
13. Instead, the Bank has a system with
a corridor that is usually ten times wider than above, and various
bells and whistles to try to keep the price within this corridor.
Each commercial bank has a reserve account, and can also deal
on a separate account via open market operations (which are neither
open nor at a market price). The Bank of England's February 2007
provides the following explanation of its "reserves-averaging
UK banks and building societies that are members
of the scheme undertake to hold target balances (reserves) at
the Bank on average over maintenance periods running from one
MPC decision date until the next. If a member's average balance
is within a range around their target, the balance is remunerated
at the official Bank Rate.
14. The Bank's most recent Quarterly Bulletin
says that aggregate reserves targets are about £17 billion.
Since the banks are required to keep this £17 billion on
deposit with the Bank, the central bank will have to lend this
money to the same banks, as well as lending the £28 billion
explained earlier. Again from the Red Book:
Open Market Operations (OMOs) are used to provide
to the banking system the amount of central bank money needed
to enable reserves scheme members, in aggregate, to achieve their
reserves targets. OMOs comprise short-term repos at the official
Bank Rate, long-term repos at market rates determined in variable-rate
tenders, and outright purchases of high-quality bonds.
15. But, rightly, the Bank understands that
the two features above, between them, would not be sufficient.
For example, one private-sector bank might be able to borrow so
much money from counterparties that, on the last day of the reserves
period, it becomes a de facto monopoly supplier of funds,
presumably at an extravagant rate. Accordingly, the Bank maintains
a corridor of "standing facilities":
Standing deposit and (collateralised) lending
facilities are available to eligible UK banks and building societies.
They may be used on demand. In normal circumstances they carry
a penalty, relative to the official Bank Rate, of +/-25 basis
points on the final day of the monthly reserves maintenance period,
and of +/-100 basis points on all other days.
16. These standing facilities resemble the
proposal above, albeit with some changes. First, these facilities
need to be invoked actively, rather than being applied automatically
to end-of-day balances. Second, because the Bank endeavours to
supply the correct amount of central bank money via the three
types of open market operations, these standing facilities are
centred on the policy rate. Third, the band is very wide, on most
days being ±100 basis points around the policy rate. Even
on the last day it is still a "penalty" of ±25
basis points away from policy.
17. This complication is a mess, and a mess
18. The complication caused problems before
the credit crisis, albeit problems less widely reported than those
that came later. Too few funds were taken at the end-June OMO,
causing inter-bank interest rates to trade abnormally high relative
to the policy rate in the following week.
19. This sudden increase in inter-bank interest
rates could not have happened with a simpler arrangement, lending
day-by-day against collateral, at the policy rate, as much as
each counterparty wants, subject to some huge counterparty-by-counterparty
limit. But the complexity of the Bank's system caused a coordination
problem: individual actions by counterparties failed to solve
the collective problem.
20. But the system's complications caused
worse problems not mentioned in the Quarterly Bulletin.
In August the rarely used standing facilities were used by Barclays,
reported by the Daily Mail on the 31st:
The financial health of Barclays was being questioned
after it was forced to ask for huge loans to bail it out.
The lender has tapped an emergency credit facility
with the Bank of England twice in 10 days.
The loans for almost £2billion are fuelling
fears that it has become too deeply involved in high-risk debt
21. For the last decade the Bank of England
has been very keen to signal to a wide range of economic agents
that future in?ation will be close to its target. Given this care
in monetary signalling, why does it allow this careless signal
about financial instability? It is impossible to know whether
this smoke helped give the impression of fire in the banking system,
and hence the run on Northern Rock, but why has the Bank created
a system that needlessly gives this false and dangerous signal?
22. The problem is that if a feature is
little used, and is then used during any sort of market turbulence,
some will believe that use of that feature contains signi?cant
negative information. The solution is a simpler system that actually
works, all parts of which are often used.
23. Next we come to the separate issue of
what collateral a central bank should accept against its monetary
policy operations. (Financial stability operations are different:
a central bank might have to lend against poor collateral, or
even against no collateral. But that shouldn't be happening whilst
the borrower's previous management are still on the payroll.)
Unlike the question of how to implement monetary policy given
a definition of collateral, the optimal definition of collateral
for monetary-policy operations is a genuinely difficult problem.
24. If a central bank lends against a wide
range of collateral, and allows counterparties to manufacture
collateral, it risks exacerbating a mis-pricing of risk. Also
the central bank would be taking far greater credit risk. For
example, consider a commercial bank that has lent money to households
in the form of mortgages. These mortgages are bundled into securities.
If the central bank were to accept these securities as collateral,
this would mean the following.
24a If the households should start defaulting
in significant numbers, the counterparty would be insolvent just
as the collateral provided by the counterparty is dropping in
pricea credit loss for the central bank.
24b Because the central bank is holding the
credit risk, in some sense it is doing the lending to the households.
If the interest rate does not re?ect the creditworthiness of the
households, how is the central bank to know?
25. Alternatively the central bank could
lend only against the best collateral: bonds issued by the government
of the jurisdiction of the central bankfor the Bank of
26. Before the recent loosening of the rules,
the Bank had a mixed stance.
26a The Bank did not allow counterparties
to manufacture collateral, as acceptable collateral are securities
issued by European Economic Area central governments and central
banks and major international institutions.
26b However, the Bank's collateral policy
implicitly entailed and entails substantial credit riskwithout
admitting that it does so. This needs explanation: the Bank accepts
as collateral euro-denominated paper. Imagine that Deutsche Bank
has borrowed some £20 billion from the Bank of England (an
unremarkable thing, as Deutsche is a major player in £ markets),
and has provided as collateral euro-denominated bonds issued by
Germany. Then, for some reason, Deutsche goes bust. In theory
the Bank of England would sell the German Bunds for euro cash,
and then sell the for £ in the foreign exchange market.
In theory. But in practice, central bank politesse would
ensure that the Bank would not be willing to be seen to be selling
the government bonds and the currency of a country whose largest
bank had just defaulted. (Consider the reverse: Barclays goes
under, and the following day the Bundesbank is seen selling gilts
and selling £how not to make friends.) So the Bank
of England would have to wait for at least several weeks, and
more likely several months, before converting the collateral into
the currency of the loan. And, in the three months following Deutsche's
default, how far might the euro fall? It might lose 5% to 10%;
and it might tumble more than 15%. Certainly very plausible is
a -10% move in the euro, which would cost the Bank of England
some £2 billion (more than the Bank's £1.86 billion
equity reported in the 2007 accounts). Hence the Bank of England,
by allowing euro-denominated collateral, is taking substantial
27. This is not necessarily a bad policy,
but the Bank should not be pretending to itself or to others that
its collateral policy protects it from credit risk. Indeed, the
Treasury Committee might like to ask the Bank to explain the purpose
of taking collateral, and judge this credit risk against that
2 Consider the difference between the policy rate and
£ repo fixing, using maturities of 1 day, 1, 2 and 3 weeks,
and 1 month. Score the standard deviation of this difference.
Consider the same measure in the non-£ currency, for
using the repo fixing (because it exists), and for the other currencies
the Libor fixings or local equivalent. If £ is worse in each
of the five maturities for which non-£ data exists, then
£ is deemed worse. Back
This closely follows the text of The pretend market for money,
originally appearing in the August 2007 edition of the journal
of Central Banking, and also available via www.jdawiseman.com/papers/finmkts/implementing-monetary-policy.html. Back
Described on page 356 of the 2007 Q3 Quarterly Bulletin under
the heading "June-August maintenance periods": www.bankofengland.co.uk/publications/quarterlybulletin/mo07aug.pdf. Back