CHAPTER 4: TECHNICAL ISSUES ON THE
IMPLEMENTATION OF POLICY
41. This chapter looks at how the MPC implements
monetary policy, and assesses how successful it has been in achieving
the objectives it has been set. We begin by looking at the inflation
target. We then discuss the time-scale for monetary policy and
the inflation forecasts.
The 2.5 per cent symmetric target
42. As noted in Chapter 2, although not stated
explicitly in the Bank of England Act, the Bank's inflation remit
has been implemented in the form of a symmetric inflation target:
inflation within one percentage point either side of a central
target of 2.5 per cent. The symmetry is in order to avoid a too
conservative monetary policy. In practice, the Bank of England
has more often undershot the 2.5 per cent target than overshot
it, whilst its inflation forecast has tended in the last two years
to exceed the target. We probed with MPC members and other
witnesses whether the MPC was in fact operating a symmetrical
target or whether they were indeed happier, as it might appear,
to undershoot rather than overshoot.
43. The members of the MPC confirmed that the
target was symmetrical. Professor Charles Goodhart told
us "we have consistently remained symmetrical about trying
to achieve, over the horizon to which we give most weight, that
it should be exactly 2.5, neither an iota above, if we can help
it nor an iota below" (Q 13). Mervyn King spoke
of "the importance of the symmetric nature of the inflation
target" and said that, while there were always shocks which
moved the committee away from the target, they were always trying
to get back onto the target (Q 151). The Governor said that
it was the tendency of all nine individuals to "recognise
the fact that the target is symmetrical" (Q 152). Professor Stephen Nickell
thought that the "symmetry of our remit is incredibly important",
not least because it facilitated explanation of why interest rates
were lowered. The symmetry provided a clear way of focussing on
the work of the MPC. Mervyn King added that adopting an inflation
target ensured the remit was absolutely clear both internally
and to the world and to the financial market and was a helpful
tool in explaining what the MPC was trying to do and thus gaining
credibility (QQ 7478).
44. For the Treasury, Sir Andrew Turnbull
and Ed Balls confirmed that the target was a point objective
of 2.5 per cent. Ed Balls was quite clear that although the
Governor is required to write to the Chancellor if inflation goes
above 3.5 per cent or below 1.5 per cent, "this does not
define a range" for inflation (Q 355). Ed Balls
suggested that it was an important consequence of the symmetric
target that deviations below the target were taken just as seriously
as those above. This would have the effect of removing deflationary
bias from the MPC's decisions and consequently encouraging a longer-term
view (Q 283). The Chancellor told us that avoiding inflation
significantly below the target was a sign of commitment to growth
(Q 1188). Governor Brash too thought "there is
a great deal of merit" in having any central bank as keen
to avoid deflation as inflation (Q 419). As already noted,
Paul Ormerod expressed a preference for just having a range for
inflation, from zero to four or five per cent and ignoring month
by month fluctuations (Q 707). He was, however, alone in
45. In practice, inflation has more often
been below the target than above it. If symmetry is important,
and the aim is to hit the target, the fact that in practice the
outcome appears to be a downwards bias needs further investigation.
There appear to be two possibilities: either the MPC is in
fact operating an asymmetric target, or the undershoot is due
to circumstances outside the control of the MPC. We accept that,
as the MPC members themselves have told us, they are working on
a symmetric target. We also accept the Bank of England's explanation
that this has been because of a need to deal with a number of
transitory effects, of which we consider a prime example to be
the fact, as pointed out to us by Gus O'Donnell, that the
exchange rate has been firmer than the Bank had forecast (Q 359).
A symmetric target in our view not only allows the MPC stability
in its operations but also, as suggested by the Bank of England
Commission encourages transparency, by giving the MPC an unambiguous
goal; and supports accountability by requiring an explanation
from the Committee if inflation strays too far from the target.
We accordingly recommend further study, perhaps by the new
Economic Affairs Committee, of why inflation has more often been
below the target than above it. Such investigation could also
examine in detail the reasons why the MPC has been so successful
in keeping inflation down; whether there are lessons to be learned
in case economic conditions change for the worse in the future;
and whether there are any issues of significance arising from
the transmission mechanism. Such a study could also examine how
the MPC takes account of the Government's wider economic policy
(including whether the "subject to that" proviso in
section 11 of the Act in fact has any significance).
The Time-scale for Inflation Control and the Inflation/Output
46. The remit is to hit the inflation target
always, and there is an obligation to stay within one percentage
point of the central target in the sense that, if the end points
are breached, a letter must be written to explain why. Nevertheless,
differences in how to interpret the remit of the MPC have emerged
amongst its members. Some, including the Governor, interpret it
as meaning that inflation should always be within one percentage
point of the 2.5 per cent target and that a letter to the Chancellor
is required if it is not. Other members of the MPC, noting that
it takes about two years for changes in interest rates to be fully
transmitted to inflation, interpret the remit as aiming to control
inflation over a horizon of two years. We say more on forecasting
in Chapter 5.
47. Gus O'Donnell said that the use of the
two-year period did not mean that a change in interest rates now
would mean that in two years time "boom-something happens":
the position could better be described as "when we change
an interest rate we know on average over the past the path that
has had to affecting inflation and we then tend to look at the
effect coming through all of it by about the end of two years"
(Q 290). Governor Brash said that no one could be very
confident in looking two years ahead, as anything could happen,
but there was no alternative to proceeding in such a way (QQ 4289).
Mervyn King told us that a demand shock would tend to push
inflation and output growth up in the same direction, whereas
a supply shock would push them in different directions. The two-year
horizon was therefore important to allow the MPC to look ahead
and see if short-term inflation would go away (Q 822).
48. The point to be emphasised here is that a
lot can happen in a two-year horizon. For example, in the next
two years, an election is unavoidable. The outcome and consequences
of this are something that markets will try to anticipate and
evaluate. It is also possible that, two years from the publication
of this report, a referendum on the United Kingdom joining the
euro may have been held or be about to be held. It follows that
unless short-term interest rates and inflation converge to euro
levels, the possibility of the United Kingdom joining the euro
would have an effect on the MPC's decisions over that two-year
period. However, the CBI told us they do not see
any need to change the 2.5 per cent target against the possibility
of euro entry (Q 255). The Governor told us that the MPC
would not speculate on the possibility of euro entry although
the MPC would have to take into account market developments arising
from, for example, people's expectations of the exchange rate
(QQ 13301335). The possibility of joining would be
treated like any other short-term external shock (QQ 13589).
49. This is not a report about the euro and we
are not intending to comment further on this possibility. We do,
however, draw it to the attention of our successor Committee.
The question of how the behaviour of the economy will change
after the general election, should a party committed to a referendum
on joining the euro win, will be a significant one for our successor
Committee to consider over the coming years. The reaction
in the markets and market speculation to the various possibilities
will also be of importance. We note the possibility of serious
effects should the markets begin to anticipate that the UK will
join the euro.
50. Another very important point that arises
from the time-scale for inflation control is often ignored. Despite
the requirement of the Bank of England Act that, subject to achieving
the inflation target, monetary policy should support Government
policy on growth and employment, we have already noted that there
is little scope for the Bank to do this. We cannot discern any
examples where this has so far occurred. One way in which the
Bank could do so, however, relates to the letter it must write
to the Chancellor when inflation is more than one percentage point
from the central 2.5 per cent target. If a shock occurs to inflation
that is forecast to breach the limits, the Bank may have a choice
to make. It could attempt to keep inflation close to the target
no matter how large the required change in interest rates, and
no matter how much that might affect growth and employment. Or,
on the grounds that the cost in terms of growth and employment
would be too high, it could choose to send the letter. In effect,
therefore, the shocks to inflation would be passed on to other
variables so that they do not affect inflation. So far this situation
has not arisen, but there is a non-negligible chance that it could.
The Bank's remit is not clear on what the Bank's decision should
be in such a case.
51. The situation is further exacerbated by the
dynamics of the transmission mechanism of monetary policy which
take time to work themselves out fully. Initially the impact of
a change in interest rates on inflation is small, and it is generally
assumed that it takes about two years for the virtually full effect
to be transmitted. As a result of the effect on inflation of changes
in interest rates being so small in the short term, the interest
rate change might need to be all the larger to keep inflation
very closely on track. Moreover, as shown in our previous Report,
it would be mainly the exchange rate channel through which the
interest rate change would initially work. This is because the
exchange rate, being an asset-price, is very volatile. In the
first instance, therefore, the impact of anti-inflation policy
would fall much more on tradeable goods sectors of the economy
than other sectors.
52. It remains unclear whether all members of
the MPC believe that it is possible or desirable to try to keep
inflation within the symmetric target it has been given. Further,
although, as the Chancellor says and is widely agreed by witnesses,
there may not be a long-run trade-off between inflation and output,
our conclusion is that keeping strict control over inflation
at all times is likely to create a short-term trade-off. We also
note that, if it takes time for interest rates to affect inflation,
any letter to the Chancellor might be better sent in anticipation
of breach of the limits, instead of after the event.
What is the Appropriate Index to Measure Inflation?
53. In our previous report, we analysed at some
length various indices used to measure inflation.
The MPC targets RPIX, which is the rate of retail price inflation
excluding mortgage interest payments. Several other central banks
that conduct monetary policy by targeting inflation use a similar
measure, but many use a different target. While there is general
agreement that it is desirable to stabilise a broad measure of
inflation, such as consumer prices, the differences between central
banks arise largely from the components that were excluded from
this measure and to a lesser degree the method of construction
of the index.
54. There are a number of reasons why certain
components might be excluded from the measure, even though it
is generally agreed that a broad measure should in fact be used.
RPIX, for example, excludes interest rate payments as it is agreed
that the measure of inflation should not itself be directly affected
by changes in the monetary policy instrument. Excluding the direct
effects of interest rate decisions also means that the MPC does
not itself have to assess how much of the change of inflation
is in fact due to its own actions in changing interest rates.
It could, however, also be argued that excluding mortgage interest
payments, for example, gives a distorted cost of living, given
the large part that such payments play in many household budgets.
Furthermore, the argument for excluding mortgage interest payments
equally applies to other interest payments, such as credit cards.
A case could also be made for excluding the direct inflationary
impact of other Government measures such higher indirect or consumption
taxes. These are in fact excluded from the index RPIY.
55. The Bank of England's Annual Inflation Report
charts the behaviour of inflation as measured by three indices
(RPI, RPIX and RPIY) and these are often very different. In any
broad measure of inflation, the individual components are likely
to behave differently from each other, which can make interpretation
of broad measures problematic and complicate the MPC's choice
of interest rates. This is because many individual items may move
in opposite directions or have different volatilities. Some components
will vary differently over different timescales (for example
food prices are more volatile in the short term than household
goods prices) and these variations have important implications
for the control of inflation. Because of the time it takes for
interest rate changes to be fully transmitted to inflation
between 18 months and 2 years the MPC is unable to do
very much about short-term fluctuations in inflation unless, as
we have argued above, it is willing to raise interest rates sharply.
There is, therefore, a case for excluding highly volatile components
from the index targeted. If such components were to remain in
the index, the MPC would not wish to respond to such temporary
shocks, unless they had long lasting second round effects on other
components and thus on the inflation being targeted. We consider
in paragraphs 84-87 the effect of outside shocks on the inflation
rates though it should be noted here that movements in, for example,
oil prices or other imported goods are foreign in origin and beyond
the control of the Bank. Many central banks accordingly exclude
such items, and target core inflation (the rate of domestically
produced output in services) instead. The Bank of England did
stress that it controls inflation through demand management via
interest rates and it follows that it would be much easier for
the MPC to control core inflation than RPIX. There are, however,
a number of second round effects on core inflation as many of
its components are not in fact controllable by the Bank. The MPC
would still need to respond to inflationary pressure arising from
these components and indeed, as indirect tax increases are outside
of the control of the monetary authority, there might also be
a case for the measure of core inflation to be net of taxes too.
56. In our First Reportwe
compared the two different ways of constructing an index number
as a weighted average of the individual components. RPIX is an
arithmetic average. Other indices, like HICP, would be
a geometric average. The Bank's quarterly inflation report published
both these indices and explained in individual cases why they
differ. Our First Report showed that if the inflation rates of
the component items were the same then the two measures would
be identical, but if they differed the arithmetic average would
be greater than the geometric average, such differences being
most notable in periods of high inflation. Other situations that
would cause a difference between the two types of index are when
one component, such as oil price inflation, dominates the scene,
or when there is imported goods price inflation due to the depreciation
of the exchange rate, or where some items, such as food, have
strong seasonal effects. It is important to appreciate, therefore,
that the index is a measure of the underlying reality. Using a
geometric as opposed to an arithmetic mean does not change the
underlying reality. If HICP comes up with a lower number than
RPIX, it does not follow that switching to the former means that
there is less inflation in the economy.
57. Professor Charles Goodhart told
us that it was his view that it would be "wrong-headed simply
because there are technical difficulties" with RPIX to shift
away from it. Professor Willem Buiter confirmed that
one would not change the index, or even the numerical target for
the index, unless there were very good and exceptional reasons
for doing so, because without a clearly defined major event to
justify the change, "the whole process [would] look suspicious"
(QQ 7580). Dr Sushil Wadhwani noted that
biases in the RPI were "an important problem" but would
only be significant if they were growing significantly over time.
He suggested that the Office of National Statistics should give
priority to doing more work on this question, a view with which
we strongly agree (QQ 6012).
58. There are, however, arguments for using a
different measure of inflation from RPIX. We are not clear how
much research the Bank of England have undertaken on this issue.
Our conclusion, however, is that, even if it were to emerge
from any such investigation that a change from RPIX was warranted
on technical grounds, there remains a strong argument for not
making a change at this stage. It is crucial to the MPC's
work in delivering low inflation that inflation expectations remain
low. A significant element of this is public acceptance of the
choice of the inflation measure to target. If there were to be
a change, any switch would run the risk of creating a loss of
public confidence in monetary policy, unless the public could
be persuaded to understand the merits of the new system. It is
our view that the dangers of any such move significantly outweigh
the technical merits of moving to a different inflation measure
at this time.
A Monetary Conditions Index
59. An alternative to targeting inflation is
to target an index consisting of several different items such
as inflation, the exchange rate, money growth and interest rates.
This is known as a monetary conditions index and has been used
in Canada and New Zealand. We do not wish to go into too much
detail on what is fundamentally a technical question. Governor Brash
explained to us in some detail how a monetary conditions index
works in New Zealand: his answer is printed in full in Q 416.
We note however that both Canada and New Zealand have now abandoned
the use of such an index as it tended to produce counter-productive
interest rate responses to exchange rate shocks.
60. The CBI explained to us the arguments for
using a monetary conditions index in the United Kingdom, an index
which would take into account not only exchange rates but also
other asset prices such as housing and equity. They cited three
possible reasons for moving to such an index:
· Such an index
would contain useful information about broad inflationary conditions
and this would in turn help to pin-point the future trend in the
narrower target and measure of consumer price inflation.
· Monetary policy
should target broad inflationary conditions, rather than just
a particular measure of consumer prices because stability of the
exchange rate and equity prices, for example, along with stability
and the price of goods and services should also be seen as goals
for monetary policy in their own right.
· Monetary policy
should be concerned with stability in the real economy as well
as price stability and also with targeting asset prices, for example
by using interest rates to burst a share-price bubble.
61. The CBI concluded, however, that even
if the MPC could devise a monetary conditions index that allowed
them to achieve greater stability in the real economy without
upsetting the primary objective of consumer price stability, the
primary objective of policy should remain as is, for the reasons
of clarity, credibility and public understanding (pp 434).
62. In our previous report, we said that there
is not one correct way to measure the price level or inflation.
That is still our view. If the index targeted has more volatile
elements that do not offset each other, there may be a temptation
to ignore them when setting policy on the grounds that their effects
on the index are transitory. The alternative, which has much the
same consequence, is to target an index that excludes them in
the first place. More research should be undertaken on these issues.
In the meantime, in the absence of an overwhelming case for a
particular alternative, and for reasons of public confidence and
given the role of the RPI in various contracts and in index-linking,
we re-iterate our view that the MPC should continue to target
63. Since the interest rate has effects over
approximately a two-year horizon, the MPC needs to take a view
of what is likely to happen over that period. This involves at
least two aspects; a forecast of what is likely to happen if there
is no change, and a forecast of the consequences of any proposed
change. This is also complicated by the knowledge that policy
itself can change month to month.
64. Mervyn King told us that the MPC needs to
forecast inflation because of the period of up to two years it
takes for interest rate changes to be fully transmitted to inflation
(QQ 7467). In an attempt to quantify this, in our
we showed that, based on the Bank of England's model, after one
year only half the full response of inflation had occurred. We
also showed that 80% of this was via the exchange rate channel
and that the longer the interest differential with other countries
is expected to last, the greater would be the initial impact on
the exchange rate. As a result of the need to look ahead the MPC
devotes a large share of its resources to inflation forecasting.
Our previous estimates also show what a critical role expectations
of future interest rates and the exchange rate have in the inflation
forecast. This is an area where we have some concern over what
the MPC does.
65. A second reason why forecasting is important
is because of the range of matters the MPC takes into account.
These are reflected in the Minutes which, before discussing the
implications of the latest data that have led the MPC to reach
an immediate policy decision, normally set out a whole catalogue
of issues arising across a range of economic factors. These are
usually in a standard format and include demand, output, prices
and costs, prices of assets and money, labour market conditions,
world economic conditions and the impact of any recent or forthcoming
Budget. Interest in these factors arises from what they might
reveal about inflation, as the Bank of England Act does not include
them as targets in their own right.
66. We asked MPC members to explain how they
reached an amalgam of all these factors to focus on the target
rate of inflation over the next two years and to arrive at a decision
on short-term interest rates. Professor Charles Goodhart
told us that, once a quarter, they have a forecasting session
which aimed "to take account of all these factors in a framework
of a coherent model" (Q 40). This raises the question
of the differences that must inevitably arise when so many distinguished
people such as the nine MPC members come together, but first we
probe a little more what the forecast actually is.
Inflation forecasts or projections?
67. We note that there is an important distinction
to draw between a forecast and a projection. A forecast is an
attempt to predict the actual outcome. A projection is a forecast
constrained to satisfy certain assumptions. If the constraints
were not satisfied then the projection would differ from the forecast.
Another way of expressing the difference is that between an unconditional
and conditional forecast. The MPC forms projections of future
inflation based on the assumption that the official interest rate
will remain constant and possibly on assumptions about unchanged
future policy in other areas such as fiscal policy. It also makes
assumptions about economic variables such as the exchange rate
and the world economic outlook. Since, in the past, interest rates
have always been changed at least once a year, projections of
inflation two years ahead based on an assumption of a constant
interest rate must be expected to differ from the inflation outturn.
Unfortunately, the distinction between forecasts and projections
is often lost in what the MPC says. For example, the Inflation
which describes the difference between the two, refers to the
forecast process. Members of the MPC in their evidence also speak
of the inflation forecast. Whatever word is used, it is important
to keep in mind that it is in fact a projection, or forecast conditional
on certain assumptions.
68. We inquired into how this forecast was made.
It is a quarterly forecast and is described in some detail in
the August 2000 Inflation Report. The following description is
given. The projection is based on a number of uncertain assumptions
on which members of the MPC may hold different views. After identifying
the main economic developments since the previous Inflation Report,
the MPC reviews the individual assumptions in turn. A central
projection is built up in the light of these assumptions and represents
the best collective judgement, the centre of gravity of opinion
on the Committee. A fan chart is then constructed to illustrate
the Committee's collective assessment of the degree of uncertainty
the variance of the distribution of the likely outcomes
and the balance or risks or skew. An outcome in the tail
of the chart is to be regarded as very unlikely. Although the
central projection is revised each quarter, the variance is estimated
on the basis of the past ten years of forecast errors. It would
appear from this description that the central forecast that emerges
is one that the whole Committee is required to sign up to, even
though they may differ in the assessment of some of the assumptions
69. Charlie Bean (appointed as an Executive
Director of the Bank, and as a member of the MPC in October 2000)
said that "the forecast represents the centre of gravity
for the Committee but everybody does not necessarily sign up exactly
to the forecast, and that is the whole purpose of Table 6B of
the Inflation Report, to indicate the extent to which different
individuals might diverge from that" (Q 1319). The Governor
said that the published forecast is "the centre of gravity,
the collective best judgement", but an individual member
may differ a little from this on the basis of their assessment
of the particular factors: "Those things are reflected in
Table 6B" (Q 1321). Mervyn King confirmed that
this was "ambitious undertaking, which no other Central Bank
does with a Committee". There were two good reasons for proceeding
in this way: transparency and allowing an explanation of interest
rate decisions to be given in terms of the outlook for inflation.
Mechanisms existed to deal with individuals' unwillingness to
sign up to the forecast (Q 803). Dr Sushil Wadhwani
thought disagreements were "healthy and entirely normal"
70. Professor Willem Buiter stressed
that they tried to predict a whole range of factors, such as world
trade, world prices and the price of oil, as well as the actions
of the US Federal Reserve and the European Central Bank, because
each such "is part of the environment within which we operate.
It is like the weather. We try to predict it presumably because
it influences our environment but that is about it" (QQ 9698).
A forward view was taken of all these factors in the external
environment and judgement sometimes overrode even market yields
(QQ 99103) although John Vickers described what went
on as not betting but rather "trying to think through what
the real forces are in the world economy" (QQ 104106).
71. Another key factor in the forecast is the
MPC's analysis of the state of the labour markets, and of individuals'
expectations for inflation. Professor Stephen Nickell
told us that shocks from the labour market did affect monetary
policy. Upward pressure on wages could either lead to continuing
inflation, if the monetary policy response to it was relaxed,
or to only a short-term inflationary pressure if there was a relatively
tight response. The labour market remained pretty tight so signs
of inflationary pressure would be taken very seriously. Current
expectations for wages seemed to be assuming that inflation would
remain around 2.5% (partly due to the work of the MPC in stabilising
inflationary expectations) but if the labour market were to tighten
further and employers began to compete for employees huge wage
increase could follow (Q 794).
72. The uncertainty surrounding the forecast
has led to various suggestions: that it should not be published;
that it should be solely a staff forecast; that it should be an
independent forecast; that each member should make their own forecast.
The Bank of England Commission
has recommended that responsibility for the inflation forecast
be taken by those MPC members who are employees of the Bank. The
Commission's Report argues that transparency will be improved
if there is "open publication of any conflicting projections
upon which the external members of the MPC have relied in coming
to their voting decision".
A staff forecast as such, however, has been rejected by the Bank
because it would create divisions between the internal and external
members; because the markets would not treat it as a forecast
of the way in which the MPC itself was thinking; and because of
sensitivities about publishing it (QQ 603, 802). There are other
arguments against each MPC member making their own forecast. First,
some individuals would find it easier to make forecasts than others.
Second, there would be speculation about the reasons for differences.
Third, presentation and communication with the public would be
weakened if a general message of agreement on the broad outlook
was complicated by the publication of "small technical differences
among the Committee" (Q801). Finally, the notion that the
MPC works as a team would be lost. While we can see some force
in these arguments, however, we consider that the case is still
open. We recommend in paragraph 94 below that the new Economic
Affairs Committee considers the alternatives to the present system
73. In commenting on the fact that members of
the MPC may have the same broad outlook for inflation but very
different views on the choice of interest rate, Mervyn King
told us that "It is the level of interest rates that matters.
I think those differences [in interest rates] will show up in
small differences in the forecasts but the differences between
members of the Committee and their outlook for inflation are tiny
in comparison with the uncertainties about the outlook for inflation
around the central projection that we might all share. Those uncertainties
are very much bigger than the sorts of differences between members
that one might find" (Q 805). Mervyn King's comments
prompt us to wonder how the MPC can have such a consensus on the
interest rate when there is so much uncertainty about future inflation.
One reason why this might be possible is if interest rates have
little effect on inflation. But this would then undermine the
whole rationale for using interest rates to control inflation.
How good is the forecasting record of the MPC?
74. Mervyn King also cautioned against thinking
of forecasts as estimates, which can be either right or wrong.
There was, he said, a sense in which no forecast can be right,
as unexpected shocks prevent you predicting the future. Hence
a forecast is not a number, and the fan charts, which were presented
as a contribution to transparency and public discussion, tried
to show that it was "the balance of risks_the uncertainty
about inflation" which determined policy. Hence an analysis
of the forecasting record of the MPC would not take the form of
assessing whether the MPC had got things right or wrong but of
professional judgement of where the balance of risk lay. The MPC
accordingly produced an annual review of its track record (Q 800).
Whilst we accept that the MPC's inflation forecasts should not
be judged purely as point estimates, this is not the same as accepting
that a class of forecasts cannot be assessed as right or wrong.
An example would be a systematic tendency to over-predict or under-predict
inflation (or output, for that matter). Another relevant consideration
is the speed at which the MPC identifies a change in the inflation
trend or the broader economic environment. This has relevance
for our analysis of the MPC's forecasting record below.
75. Much of the error in forecasting inflation
has been attributed to the exchange rate. As discussed in the
our First Report, the MPC has in the past used uncovered interest
parity (UIP) to forecast the exchange rate according to which
the exchange rate is determined by interest differentials. In
Appendix 4 of that Report we discussed in detail how difficult
it is to forecast the exchange rate with any certainty using this
approach, even though it may well be the correct way to do so.
Given the importance of the exchange rate in forecasting and controlling
inflation it is worth adding that in a recent paper Dr Sushil Wadhwani,
then a new member of the MPC, has expressed dissatisfaction with
this approach to forecasting exchange rates.
Several major problems with this method were also examined in
our First Report.
The assumption of a constant interest rate for the projections
can only be expected to add to the role of the exchange rate in
causing errors in the inflation forecast.
76. A detailed assessment of the MPC's inflation
forecasting record is published in the Bank's August 2000 Inflation
The forecasts are for one and two years ahead. The assessment
period begins with August 1997; the last available for one year
ahead was made in May 1999. Charts A and B show that there has
been a persistent tendency for the projections to exceed the outturn
of inflation. According to the Inflation Report, the one-year-ahead
projections of the mean made between August 1997 and May 1999
were on average 0.3 percentage points above the inflation outturn.
The two-year-ahead projections of the mean made between August
1997 and May 1998 were on average 0.7 percentage points above
the inflation outturn. The Inflation Report attributes these forecast
errors mainly to the exchange rate being stronger than was assumed.
It is also noted that interest rates tended to be lower than assumed
in the projection. This was because the MPC responded to weakening
external pressure on inflation (caused by lower world growth rates)
by reducing interest rates. Chart D shows that the MPC's one-year-ahead
output projections made between August 1998 and May 1999 were
persistently below the outturn, sometimes by as much as 1.5 percentage
points. Thus the record shows that more recently the MPC has been
persistently too pessimistic about both future inflation and output.
77. The Governor said that he was "rather
proud of our forecasting record", and that it had been "remarkably
accurate". He added that "if you can produce a forecast
which has got a better track record than that, I shall be absolutely
delighted". He also said that the persistence in the errors
was the result of looking two years ahead, and that there were
still only a few observations on which to make a judgement (Q 1307).
Charlie Bean told the Committee that the forecast performance
had been better than expected, that the errors had been small
and that the inflation performance had been surprisingly good.
He attributed the output forecast errors in part to the unexpected
strength of sterling (QQ 13071308).
78. We tried to probe further with individual
members the case for more self-analysis of the forecast. Professor Willem Buiter
said that the most unexpected factor had been the exchange rate
and as he could not explain after the event why sterling was where
it was, the fact that he had not predicted it correctly in advance
was not too serious (Q 14). Professor Charles Goodhart
took comfort from the fact that the Bank had not had to write
a letter of explanation, even though statistically one might have
expected that two such letters would have had to have been written,
while John Vickers welcomed the idea of periodic review (Q 15).
Dr Sushil Wadwhani told us that the staff regularly
present the MPC with a comparison of their forecasts against external
forecasts, both showing what such forecasts were saying and trying
to explain why there were differences (Q 606).
Uncertainties surrounding the inflation forecast
79. In his evidence, Mervyn King stressed
that the MPC's forecasts of inflation were subject to considerable
uncertainty (Q 800). This is the reason why the MPC publishes
its fan charts. The shading around the central projection decreases
further away from the centre to reflect the lower probability
the MPC attaches to inflation taking such extreme values. The
bands widen the further ahead the projection. For two years ahead
the band-width is on average about two percentage pointsabout
one percentage point either side of the central projection. This
implies there is about a 10% probability that inflation will be
one percentage point above or below the central projection. The
corresponding figure for one year ahead is about 0.75 of a percentage
80. We accept entirely Mervyn King's emphasis
on uncertainty and acknowledge the attempt to represent this in
the fan charts. Nonetheless, it is instructive to relate the uncertainties
in projected inflation reflected in the fan charts to the forecasting
record. As noted, the mean forecast error one year ahead has been
in the range 0.50.7 percentage points. Although this is
large in relative terms, however, we accept that these errors
are still small in absolute terms.
81. One of the factors that is said to have caused
a difference of opinion in the MPC about the interest rate decision
is the impact of technology gains on inflation. According to one
view, buoyant domestic demand is likely to have less impact on
future inflation because productivity gains have reduced unit
costs. To have a permanent, and not just a temporary, effect on
inflation, however, such productivity gains would need to be repeated
frequently. A one-time productivity improvement would have a permanent
effect on the price level, not the inflation rate.
Uncertainties surrounding the interest rate decision
82. Members of the MPC have told this Committee
that they attempt to control inflation by demand management. By
reducing aggregate demand, inflation is decreased. It is worth
observing, therefore, that the projections for output growth since
August 1998 have been repeatedly lower than the outturn - by roughly
1.5 percentage points. This is attributed to stronger domestic
demand than expected due to the lower interest rates, and to wealth
effects arising from equity capital gains. Thus at the same time
that GDP growth has been greater than expected, inflation has
been lower. This raises a strong concern about the interest rate
decision. If the MPC had forecast inflation more accurately, interest
rates would presumably have been lower than they were. But given
the relation supposed to exist between demand and inflation, if
the higher output growth had been forecast accurately then interest
rates should have been higher than they were.
83. Another puzzle about the interest rate decision
is that in the months from July to September 2000 the MPC central
inflation projection two years ahead has been about 2.75%, yet
interest rates have remained unchanged. If it takes about two
years for interest rates to have their full effect on inflation
then one might have expected an immediate increase in the interest
rate. In fact, only four of the nine members of the MPC have voted
for an increase.
Is the current approach suitable for supply shocks?
84. Since its inception the focus of the MPC
has been on the control of inflation by demand management through
interest rate policy. As long as only demand shocks are affecting
inflation this will almost certainly remain a viable approach.
We are also concerned by how the Bank responds to supply shocks,
such as imported inflation due to, for example, oil price shocks,
and the gains to inflation due to technological improvements that
reduce unit costs. The evidence we have received appears to
indicate that supply shocks would be treated in the same way as
demand shocks, but we are not convinced that this would necessarily
be appropriate. This is one area where, to date, the environment
has not been too difficult for the MPC, especially when compared
with the severe problems of the past. The current approach might
be sorely tested in less favourable situations.
85. The Bank of England Commission suggested
that "in the case of a large systemic shock, it might be
inappropriate for the Chancellor to allow the MPC to continue
setting interest rates designed to restore the level of inflation
compatible with the pre-existing target range". They suggest
that, under such circumstances, the Chancellor announces a path
of targets and ranges leading back towards low inflation at a
speed he judged appropriate.
In New Zealand, however, Governor Brash is directed to ignore
the immediate effect of supply shocks, at least in the first round
of inflationary consideration, although there would be a tightening
of policy if, for example, higher oil prices tended to lead to
higher wages, (QQ 4246).
86. We asked our witnesses about a number of
different supply shocks. As far as oil prices were concerned,
John Vickers told us that they were not a special factor
(Q 33) whereas Professor Willem Buiter said that
they remained "a big component" because they were set
by a cabal. Mervyn King was more dismissive, however, of
the importance of equity prices, as these could not be predicted
anyway. He questioned whether there was such a thing as a "bubble"
in terms of the level of asset prices but he did note that central
banks had to be modest in forming a judgement on whether a valuation
was right or wrong. The MPC would of course look at any risks
arising from asset prices (QQ 8136). The Governor
told us that the significance of the housing market, too, was
over-exaggerated and the MPC did not take much account of it (QQ 205, 213).
Chris Allsopp said that asset prices were indeed less important
than looking at the overall picture (QQ 809810). Professor Stephen Nickell
told us that there had also been fewer shocks arising from the
labour market, as the structure had changed due to changes in
unemployment benefits and union activity (QQ 79091).
87. More generally, in their evidence to the
Committee, members of the MPC were less clear how they would respond
to higher inflation due to a supply shock, as opposed to a demand
shock. It remains unclear whether they would take action to keep
inflation on target, or take the view that this might be an occasion
when a letter to the Chancellor would be more appropriate. Perhaps
there is a clue in the recent increase in the oil price which
appeared for a time to be a possible threat to inflation. The
central forecast of inflation in August 2000 rose to 2.75 per
cent, but interest rates were kept unchanged. In general it may
be asked whether an interest rate increase is the right reaction
to a supply shock. Arguably, the correct response to an oil price
shock is to allow the price level to rise and then to re-base
future inflation at the higher price level. This argument reflects
the fact that demand has not increased, that the oil price shock
is already deflationary and the required response is a substitution
away from the use of oil. Notwithstanding our previous discussion
of the choice facing the MPC between keeping inflation on track
and safeguarding growth and employment when inflation is likely
to breach the one percentage point limits, this response to a
supply shock would imply temporarily higher inflation, and might
possibly necessitate a letter from the Governor to the Chancellor.
How important is forecasting for successful inflation
88. We have been told that forecasting is a pretty
uncertain process, almost random in its operation. In such a world,
it could be argued that the work of the MPC should be discontinued.
We reject this view. If the MPC is to be fully effective, it is
crucial for them to have good forecasts of future inflation in
making the current interest rate decision. The forecasts are conditional
on key assumptions about the interest rate and certain external
economic variables and members of the MPC sometimes differ in
their opinions about these. There is considerable uncertainty
about the forecast, but this uncertainty seems to be unable to
account for the much larger forecasting errors that occur in practice.
There is good reason to think that this might be due to an inappropriate
treatment of future interest rates in forming the forecast. The
exchange rate forecasts seem to be a contributory factor, and
the method of forecasting is another, and probably related,
area of concern.
89. As the MPC has been in operation for several
years, we consider that the Bank should now be in a position to
review its performance in general so far. Under section 16(1)
of the Bank of England Act, a sub-committee of the Court of Directors
of the Bank is required to keep the procedures followed by the
MPC under review. Our view is that what they should concentrate
on is forecasting performance. We note that the Bank has set out
for the Commons Treasury Committee
how they would go about assessing the MPC's forecasting performance.
We recommend that a review of the method of forecasting inflation
used by the MPC should take place, with a view to discerning whether
it is state-of-the-art. This review (in effect an audit) should
be sponsored and financed by the Court, although they themselves
would not be the body which undertook it.
90. A matter of interest to our witnesses was
whether the MPC was sufficiently alert to regional factors. The
CBI told us that there were very helpful contacts between business
and the Bank, sometimes through the Bank's Agents and sometimes
through visits by individual MPC members. Any problem was caused
not by a lack of awareness of what was happening in the regions
but by the need to set a common interest rate for the country
as a whole (QQ 2423). For the purposes of the narrowly-focused
inquiry we are conducting, we have concluded that the MPC is sufficiently
alert to regional factors. We recommend, however, that the
new Economic Affairs Committee conducts a thorough review of the
importance of regional variations in the economy and of the steps
necessary to take account of them in the formulation of policy
across the board.
How successful has the MPC been?
91. One of our witnesses, Paul Ormerod,
questioned the whole basis on which the MPC works. It was his
view that, with inflation low around the world, it would be difficult
to claim that any particular central bank had any particular responsibility
for low inflation in a particular economy. He also questioned
the assumption, on which the operational arrangements for the
MPC are based, that it is possible to predict with reasonable
accuracy future events in the economy and to exercise reasonably
precise control over outcome by actions which are taken today
(Q 706). Despite these fears, we note that to date the MPC
has been successful in achieving its inflation objectives.
92. Looking ahead, will the MPC be able to continue
to be as effective should economic conditions change? It has been
suggested by the Bank of England Commission and others that the
recent economic environment has been rather benign for the operation
of monetary policy. The Bank of England Commission noted that
many of their witnesses had pointed out that "the international
inflation background_has been benign_[and] that both inflation
and inflation expectations were low at the time when the new arrangement
first came into force".
93. We put this point to Mervyn King and
his view was that the MPC could take some credit for what had
happened to inflation, because of its response to the economic
shocks during its period of operations. He cited the 1997 Asian
crisis; the Russian default and devaluation in 1998; and the sharp
pick-up in world growth in 1999. He also noted that the exchange
rate had risen very sharply immediately before the establishment
of the MPC, and had then continued to rise (Q 743). We are
inclined to agree with Mr King. Saying that the economic environment
has been relatively benign does not mean that it has not been
confronted with serious problems. It certainly does not imply
that the MPC has not worked well, and has not helped to keep inflation
low and stable.
94. It is our view that this system has worked
well in the relatively benign recent economic conditions. It may
not be so easy in the future, when conditions get tougher as they
surely will, to arrive at and rely on a collective forecast. We
recommend that the Bank considers alternatives to the current
procedure for preparing the forecast, even if only to satisfy
everyone that those procedures will indeed be sufficiently robust
in a harsher economic climate. The general question of
forecasting is one that the new Economic Affairs Committee may
wish to examine.
26 See note 10 above: Paragraph 7.11. Back
See Chapter 3 and Appendix C of our First Report. Back
Paragraphs 3.17-3.30. Back
Chapter 4 of our First Report. Back
August 2000, p 49. Back
See footnote 10 above. Back
Paragraph 7.32. Back
Speech to the Senior Business Forum, Centre for Economic Performance
"The Exchange Rate and the MPC: what can we do?" (31
May 2000) available at www.bankofengland.co.uk Back
Chapter 4. Back
Pages 63-66. Back
Paragraphs 7.14-7.18. Back
Op.cit. n8 above: Appendix 12. Back
Paragraph 3.2. Back