Memorandum submitted by Professor Shelia
Dow, University of Stirling
CHANGES TO
THE PROJECTIONS
1. There have been several changes recently
in the way in which the inflation and GDP projections are constructed
and shown:
(a) The lead projections are now, since the August
Inflation Report, those based on market interest rate expectations
rather than assuming constant interest rates (see discussion below).
(b) The projections are shown, since the August
Inflation Report, for a year beyond the two-year target
horizon. This is helpful, in that the policy decision often refers
to trends beyond the two-year target horizon, about which there
is now greater transparency. For example, although a small pick-up
in inflation was projected beyond two years in the November Report,
the fact that the risks were seen to be on the downside seems
to have contributed to the decision to keep the rate on hold in
November.
(c) Market interest rate projections are now
shown, since the August Inflation Report, with their own
fan chart. These are constructed in a quite different way from
the inflation and GDP fan charts, being based on market futures
contract data, from which the risks to the central forecast are
imputed.
(d) In the meantime, the Bank have been using
their new quarterly model, BEQM, but the full details have not
been made public.MARKET RATE
EXPECTATIONS
2. When constant interest rates were assumed
throughout the forecast period, then the forecasts presumed no
monetary policy action, which is clearly unrealistic, although
it makes the immediate policy decision more clear. Allowing for
rates to change during the forecast period in line with market
expectations allows for a rate response to the path taken by inflation
and output, feeding back in such a way as to keep inflation within
the target bands of 1% and 3%. But, while this Report shows market
expectations that interest rates will be such (by the Bank's analysis)
that inflation will be spot on target in two years, this has not
always been the case. Further, the market expectations for inflation
shown in Chart A, page 46, suggest that the market on balance
expects the MPC to undershoot the target. It appears that the
MPC has earned market credibility, ie market confidence that they
will keep inflation within the target range. But either the market
has expectations at times that the MPC will not aim for the precise
target, or else the inflation which follows from particular interest
rates is different to the MPC forecasts due to different analysis.
3. While allowing for rate changes in response
to events is more realistic than constant rates, the policy component
of the projections, and thus the meaning of the projections, are
less clear. Embedded in the projections are specific assumptions
about MPC rate decisions which have not been made transparent.
If the market expects inflation to accelerate, they will also
expect the official rate to rise, which will then ensure that
inflation stays close to target. So a projection which reflects
an expectation of a rate rise should normally lead to a rate rise.
The exception, apart from unforeseen shocks, is if the market's
analysis, and thus forecasts, are less sound than those of the
MPC. It is puzzling therefore that there is not better information
about the Bank's new quarterly model, to assist the formation
of market expectations.
4. If monetary policy is based on these projections,
we end up with a reflexive process by which official rates are
set in accordance with market expectations of official rates.
Indeed, were that not the case, there would be concern that the
policy process lacked transparency, or else that the MPC's analytical
approach had changed from when the market's expectations were
formed. This is quite the reverse of thinking in terms of policy
"shocks". The question is then, not just how to understand
the projections, but also how to understand policy. Since the
forecasts are now based on estimating the market's estimation
of what the MPC will do on the basis of market expectations, how
far is the MPC implicitly effectively endorsing market expectations
on the one hand and how far trying to influence them on the other?
Is it now a matter more of influencing behaviour by influencing
expectations through other means than interest rate changes (if
the latter are generally already factored into expectations)?CHANGE
IN INFLATION
MEASURE
5. The index for the target rate of inflation
was changed from RPIX to CPI last December, and the target rate
was reduced by 0.5% from one series to the other. While earlier
Inflation Report charts had suggested that the difference
between the two series was greater than 0.5% by an average of
0.3%, a new chart shows the difference to be greater still. Chart
4.10 shows the two series from 1997 in relation to their averages
over that period. These averages are over 1% point apart over
that period. We may expect the difference to fall in the short-term
now that house prices are decelerating. Nevertheless, a 2% CPI
target is clearly looser on average than a 2.5% RPIX target.
6. This should go some way to address the charge,
as recently laid by the House of Lords Select Committee on Financial
Affairs, that the MPC have erred too much in favour of avoiding
inflation in the top half of the range. Other things being equal,
this could have been expected to persist if markets expect the
MPC to continue this pattern and their expectations have a substantial
influence on policy. Average CPI has generally been well under
2% since 1997. But if the CPI target equivalent to 2.5% RPIX is
less than 1.5% (as implied by Chart 4.10), rather than the actual
target of 2%, then the record looks less deflationary. Nevertheless,
given the evident scope for fluctuations in the RPIX/CPI differential,
and the incomplete understanding of the differences, it would
be helpful to know how the Bank have handled the change from modelling
in terms of RPIX to modelling in terms of CPI.REGIONAL
IMPLICATIONS
7. The November Report highlights the turnaround
in the housing market, which applies in all UK regions other than
Scotland. A Chart in the August Report had shown the ripple effect
of house price increases outwards to those regions where house
prices are still substantially lower. This illustrates the regional
variation in forces at work in the economy. The November Report
casts doubt on the long-term relationship between house prices
and consumption in the UK as a whole. But, for example, the RBS
Retail Sales Monitor shows retail sales running much stronger
in Scotland than the UK as a whole, mirroring the regional pattern
in house prices. This suggests that UK-level analysis could benefit
from analysing regional decompositions of national data. Where
the timing, as well as the strength, of developments differs by
region, national totals can mask a range of causal mechanisms
which would be evident in a regional decomposition. Such decomposition
would therefore help understanding at the national level. But,
to the extent that there was a better understanding at the regional
level, regional differences in the unemployment/inflation trade-off
could also be addressed, which would in turn improve the UK level
trade-off. But regional analysis is dogged by problems of data
availability for all UK regions.
17 November 2004
|
