Examination of witnesses (Questions 300-319)|
TUESDAY 27 FEBRUARY 2001
300. Presumably, what Dr DeAnne Julius has been
saying is that there are, at least, some changes in the labour
market which mean that it is safe to cut rates. You do not believe
(Professor Nickell) I believe that there have been
lots of changes in the labour market which explains why, for the
last year or so, we have been in a situation where unemployment
has been at its lowest level for a generation, and there has been
evidence of considerable excess demand in the labour market without
earnings and wage settlements taking off. Yes, I think that there
has been a big change in the labour market over the last 15 years.
The issue as far as the rate cut in December, January and February
is concernedthe issue for mewas not really a labour
market issue, except to the extent that over the period December,
January and February there was a sort of build up of evidence
that the labour market was on the turn, so to speak. Whereas previously
employment had been rising rapidly, by December the rate of growth
in employment had fallen to below the growth of the population
of working age, and by February the rate of growth of employment
had fallen to zero. So the labour market had certainly slowed
down. I think the key thing for me was that whereas in December
and January domestic demand and domestic demand prospects seemed
to be relatively buoyant, by February we had our first real piece
of evidence that domestic demand was actually slowing as it would
have to do to make room for prospective increases in expenditure
by government. So because the fourth quarter growth of GDP was
considerably lower than previous quarters, while some of it may
be a special case, there was in February real evidence of declining
GDP growth and that is the thing that tipped the balance for me.
Sir Michael Spicer
301. Governor, I wonder if I could begin by
asking you whether the MPC takes its inflation forecast at all
(Sir Edward George) I do not think we can take it
too seriously, Chairman. That is our remit.
302. You do take it seriously?
(Sir Edward George) Controlling inflation is our remit
and, therefore, the forecast is a very important part of the process.
It does not seem to me that we can actually take it too seriously.
303. It is an important part of the process
but you do not take it too seriously?
(Sir Edward George) No. I do not see how we can take
it too seriously.
304. The reason I ask is because if one looks
at the fan charts which you produce in the Inflation Reports,
first of all it compares between Inflation Reports and the fan
charts show a risk profile which, in many cases, varies very much
between forecasts. Yet, mysteriously and magically, the central
forecast always ends up on 2.5. If one takes another example,
on page 66 of the latest Inflation Report, where you unusually,
for obvious reasons, produce two inflation forecasts based upon
different interest rates, again the risk profiles compared are
very different. Again, magically, bang, we get 2.5 again. Is not
the whole thing becoming a bit spurious? It seems to me a bit
(Sir Edward George) You have to bear in mind, as far
as our own forecast is concerned, based on unchanged interest
rates, that we combine the Inflation Report process and the forecast
process with our policy decision, so that we do as we did this
time, we adjust the interest rate in order to produce a forecast
which is consistent with achieving our target. It can be either
side of 2.5 per cent, but you would actually expect that over
the kind of forecast horizon. As far as showing what would happen
under other interest rate conditions is concerned, we use the
market interest rates and that actually shows, I think, on this
occasion that the market interest rates are lower than our constant
interest rate assumption. That shows that inflation actually would
rise above 2.5 per cent in two years' time.
305. I wonder if I could ask Mr Bean, who is
responsible most directly for the forecast, about this. The Governor
has just said that you feed in, in this case, alternative different
interest rates and see what the effect of that is. Actually, you
do not, do you? One of the criticisms of the forecasting system
in terms of its credibility is that on the whole you keep the
existing interest rates you have got and feed those throughout
the system and do not make allowances for any potential changes
in interest in the future?
(Mr Bean) What is certainly true is that the projections
that we look atand I think it is important to look at these
not as forecasts of what will happen, but as conditional projectionsare
conditional on a particular path of interest rates. The forecasting
process is conducted on the assumption that interest rates stay
unchanged over the forecast horizon from where they are. Now,
to the extent that inflation at the end of the horizon is a long
way from where the target is, there is an implication that interest
rates will need to differ at some stage from the benchmark projection
we have put in. In the case, for instance, of this forecasting
round, the forecasting round was actually conducted most of the
time on the assumption of unchanged interest rates of 6 per cent.
Then, of course, they were actually changed at the February meeting
and the forecast was closed with interest rates at that new lower
level. And we do, as the Governor says, also look at what the
inflation projections would look like with the market interest
rate assumption. But what we do not have is an extensive examination
of the alternative interest rate profiles that might generate
particular inflation profiles, or particular out-turns for inflation
at the end of the horizon.
306. Mr King, I wonder if I could put the question
to you. In that case, it is absolutely inevitable, is it not,
that the forecast will be inaccurate because you are feeding in
inaccurate data. You are feeding in an assumption that interest
rates remain static over a two-year period, and that is almost
inevitably not going to be the case.
(Mr King) I think of a forecast as a probability distribution
rather than a point estimate. If you insist on thinking of it
as a number then every forecast anyone ever makes is going to
be wrong. You have to think of it as a probability distribution.
I think, as Charlie said, the key point about the process is that
we did carry out this forecast round by saying "Let's make
a conditional projection; what would happen if interest rates
were to remain unchanged at 6 per cent?" We then had a look
at the fan chart and the central projection was below the target
of 2.5 per cent. I think what matters to most members of the Committee
is the balance of risks. It is not just the central projection,
it is the entire shape of the fan chart. We then adjusted policy
so that the fan chart did indeed, in terms of the central projection,
come close to 2.5 per cent. So I think what is perhaps missing
from the perception which outsiders have is that we look at the
fan charts and the different levels of interest rates, we set
interest rates at a level such that the fan chart looks as if
the central projection is coming back to 2.5 per cent. That is
precisely what we are supposed to do, as the Governor said. In
fact, it is not always true that the central projection is 2.5
per cent. Last August we published a projection where the central
projection was above 2.5 per cent, and because of the balance
of risks, and because of the desire to see what was happening
to the labour market we explained carefully, both in the report
and in the associated press conference, why, despite the fact
the central projection was not exactly on 2.5 per cent we had
not changed interest rates. Indeed, further, we published in November
a box in the Inflation Report explaining why in general terms
one should not expect any mechanical link between the central
projection and the policy decision.
307. That is very interesting. Before I come
to that point, if I may, just picking up a point about the probability
distribution, part of what I am trying to say is that you produce
a variety of probability distributions, risk analyses and so on,
and yet the central projection, on the wholeyou have picked
on one exceptionalways seems to come back to 2.5 per cent.
(Mr King) It should come back to 2.5.
308. There seems to be a certain amount of artificiality
(Mr King) I do not think it is artificial. It is our
target. It would be most peculiar if it were the case that our
forecasts were all over the place and bore no relationship to
2.5 per cent.
309. So it is a target, not a forecast.
(Mr King) No, it is a projection under the current
level of interest rates, which you would expect, in most circumstancesthough
not allwould have the property that inflation would tend
to move back towards the target at around that horizon. As I said,
we spelled out in the box in November why there would be casesand
this happened on a number of occasions, not just last Augustwhere
the central projection could be away from 2.5 per cent. As Charlie
said, there would then be a presumption of thinking about a rate
change but not immediately, and not if the balance of risks were
heavily skewed on one side rather than the other, and not if we
were concerned about the path of inflation before two years or
after two years. All of these considerations are relevant. Perhaps
we have more explaining to do in order to show the link between
the outlook for inflation, which is the fan chart (not just the
central projection but the entire fan chart), and the policy decision.
For a long time now we have been trying to point out in speeches
or in answers to questions that there is no mechanical link between
the central projection in the fan chart and the policy decision.
310. If there is no direct relationship between
interest rates and the forecast
(Mr King) I did not say that. I said there is no mechanical
link between the central projection, which is only one part of
the fan chart, and the policy decision.
311. If there is no direct relationship between
the central projection, which is what everybody basically looks
(Mr King) I hope they will not. The whole point of
the fan chart was to get away from that. We talk about the balance
312. If they do not look at the central projection
and look at the outer rims of the fan chart, and only take the
outer rim seriously, you are saying that 4 per cent inflation
may be something which we are heading for.
(Mr King) There is always some probability of inflation
being at that level, of course.
313. Essentially, the central projection must
be the key indicator.
(Mr King) It is one part of it, and I think the balance
of risks around the central projection is the key part. Of course,
the central projection is important, but it is not the be-all
and end-all of the process.
314. But it is highly critical and important.
(Mr King) It is not a sufficient statistic for the
315. But it is still very important. To say
that there is no mechanical relationship between the central forecast
and the input of interest rates does put into some sort of question
the whole point of the exercise.
(Mr King) With great respect, I do not think it does,
Sir Michael. The point is that we are focusing here, and your
question is about, the central projection on precisely a two-year
horizon. That is a single number. Policy must reflect the entire
distribution of possible out-turns for inflation. That is what
constitutes the inflation outlook. So the balance of risks around
that, and the path of the central projection before and after
the two-year horizon, are all important ingredients in deciding
whether or not to change the policy as appropriate.
316. Given this unpredictability which you point
to and which is represented in the fan chart, given the fact that
we had a witness yesterday, Mr Kohn, from the Federal Reserve
Bank of America who said that all forecasts are inaccurate, is
it not therefore true (and perhaps I could address this to the
Governor) that what you are really doing in terms of interest
rate policy is taking very short-term factors likeI do
not knowbalancing the effect of the American economy down-turn
against tight labour markets, for example, and taking on policy
based on what you see outside the window, to a very large extent,
projecting forward in as an intuitive way as you can. That is
really the essence, and all the paperwork and the Inflation Report
concentrates, perhaps rightly, on short-term factors rather than
these, perhaps, rather arcane long-term forecasts.
(Sir Edward George) I just think that is a total misperception,
if I may say so. Of course, there are a huge number of factors
that go into reaching policy decisions. That certainly includes
short-run developments, but we are not looking at short-run developments
simply for their own sake and responding to those on a kind of
day-by-day basis, which is what very often influences financial
markets. We are actually looking at those short-run developments
for what they tell us about what is actually happening in the
economy in a longer-run sense. We have as a benchmark the forecast
that we made some time before, and we are measuring the incoming
data against that benchmark and saying how is the thing performing
in relation to what we would have expected to happen? Or what
we assumed or projected might happen? If we were to simply respond
to what we saw out of the window at any particular moment in time,
we could adjust interest rates every day. I do not think that
would be a very sensible way to run policy. What we are then doing
is actually trying to look at the kind of underlying trends and
developments in the economy; we are looking at what is happening
to the different components of demand. On some of those things
we know more than we do on other things. We have information,
for example, about what the Government intends on public expenditure.
We have a lot of past experience of past behaviour, what is affecting
consumer spending, what is affecting investment spending? It is
not a straightforward predictor of what is going to happen in
the future, but by actually looking at the behaviour in the past
you can establish relationships which show a kind of average behaviour
and you can look at the latest data to try to understand what
might be happening in the current situation, and what the risks
are on either side. So you put all of those things into the forecasting
process, trying to measure the demand side influences, and then
you have to look at what you think is happening on the supply
side too. You have to look at what is happening internationally
as well as what is happening domestically. So all of those things
have to go into the formation of a view about what you think is
the most likely thing to happen in the economy. That really is
the centre of gravity that we reflect in the fan chart and what
are the risks around that projection. All of those factors influence
our individual judgments about what we need to do to interest
rates in order to keep on track to hit the inflation target.
317. Given the unpredictability your colleagues
have pointed to, you would argue, would you, that you, as experts,
have some greater insight into the future than us mere mortals
who do just use intuition?
(Sir Edward George) I would certainly argue that as
a matter of process we have built into our policy process more
systematical analysis than people who just look out of the window,
318. That gives better answers?
(Sir Edward George) I think it reduces the risk of
bad answers. Of course, you can still get bad answers because
it is not a kind of precise science world we are living in.
319. Mr King, Members want to cover lots more
subjects, but I want to pursue Sir Michael's points in a little
more detail in a different way. If we can start with Table 6A,
which is on page 67, this shows the probability range for the
future path of inflation over the next year. According to this,
at the end of this year there is a 25 per cent probability that
inflation will be less than 1.5 per centthat is more than
1 per cent below the targetand at the other end of the
range there is a less than 1 per cent probability that it will
be over 3.5 per centthat is to say more than 1 per cent
above the inflation target. The first question I would like to
ask you is, starting from the target of 2.5 per cent, is it your
view that missing the target below by more than 1 per cent and
missing the target by above more than 1 per cent of the targetdo
each of those carry an equal cost or is one less costly than the
(Mr King) No, I think, in terms of our objectives,
we are equally concerned about deviations above or below. In terms
of policy setting, the initial starting point makes quite a big
difference, because if you start below the target there is actually
very little you can do to influence inflation over the first 6
to 12 months of the forecast horizon. So it is actually more important
to look further ahead. That is where we can influence the outcome.
So I think the fact that the probability that inflation may be
below 1.5 per cent exceeds the probability that it will be above
3.5 per cent is not directly relevant to the question of setting
policy today, when you know there is a long lag between the impact
of changing policy and its effect on inflation.