Printer Friendly

Changing the inflation target.

Since May 1997, when the Bank of England was given operational independence to set monetary policy, the Monetary Policy Committee (MPC) has been responsible for setting short-term interest rates to ensure that the Government's inflation target is met. The target is currently 2.5 per cent and the target measure is the Retail Price Index excluding mortgage interest payments (RPIX). If RPIX inflation deviates more than 1 per cent from the central target, the Governor of the Bank of England is expected to provide a written explanation to the Chancellor of the Exchequer as to why the inflation target has been missed.'

The Government has recently announced its intention to redefine the inflation target in terms of the Harmonised Index of Consumer Prices (HICP), the measure monitored by the European Central Bank (ECB), in the next Pre-Budget Report. The inflation target will remain symmetric. Unless the Government wishes to change the overall monetary stance, a central target needs to be set for the HICP measure of inflation that would be consistent with the current inflation target of 2.5 per cent for RPIX inflation. In addition, the target needs to be set so that it is consistent with the objective of helping to "ensure inflation expectations in the UK remain in line with those of the Euro Area" (HM Treasury, 2003, p. 26). The ECB aims to keep inflation, as measured by the HICP, below but close to 2 per cent over the medium term.

Differences between the RPIX and HICP measures of inflation

Chart 1 graphs both the RPIX and HICP measures of inflation since 1989. Although the two series broadly follow each other over the period as a whole there are significant differences. Most notably it is clear that, since 1989, inflation as measured by the RPIX has been above inflation as measured by HICP, apart from the period April 1991-March 1992. This period is an exception caused by the abolition of the Community Charge (the Poll Tax), which reduced RPIX inflation but did not have a similar affect on HICP inflation (O'Donoghue, 1998).


The way in which the indices are constructed and the differences that result between the two series are described in some detail elsewhere (see e.g. the monthly release 'Consumer price indices' issued by National Statistics; O'Donoghue and Wilkie, 1998; Bank of England, 2003); here we simply note the two main differences between the indices. These occur due to differences in the method of aggregation and coverage. At the most basic level, the RPIX uses arithmetic means to aggregate prices, whereas the HICP uses geometric means, and this causes the two price levels to drift apart. Since 1995 this has on average accounted for a difference between the RPIX and HICP measures of inflation of 0.49 percentage points. The most important difference in coverage, in terms of explaining the difference in the rate of inflation in the two indices, is the exclusion of owner-occupier housing costs in HICP. These are included in the RPIX in the form of housing depreciation, local authority taxes and buildings insurance. Housing costs have had a considerable impact on the difference between the two inflation measures since 1997, as this has been a period of strong growth in house prices.

Chart 2 shows the contribution of the two major factors behind the difference between the RPIX and HICP measures of inflation: the method of aggregation and housing. The acceleration in house price growth last year has caused the two indices to diverge, with the housing component accounting for a difference of 1 1/4 percentage points between the two inflation measures in May 2003.


Since 1989 the mean difference between the RPIX and HICP measures of inflation has been 0.7 percentage points. Since the MPC has targeted the RPIX measure of inflation, the mean difference has been somewhat higher at 1.0 percentage points. More recently the difference has risen above 1 1/2 percentage points. It is anticipated that the recent divergence between the two series will come to an end as we see a slowdown in the housing market.

Will the level of interest rates change?

We assume that the Treasury does not intend to change the underlying target for inflation, and hence we need to judge what the equivalent target for the HICP should be. In the longer term, a HICP inflation target set equal to the current inflation target for RPIX less the index number based drift between the RPIX and HICP measures of inflation should on average deliver broadly the same level of interest rates, all else being equal. Ignoring the more general discussion of the costs and benefits associated with the precise level of the inflation target, a target that was set higher (lower) would deliver, again all else being equal, commensurately higher (lower) nominal interest rates, leaving the real rate of interest unchanged. Thus, a HICP target of around 2 per cent is likely to be both consistent with the current framework and over the longer term to be of little consequence for the level of interest rates. However, in the shorter term, as illustrated in the previous section, the difference between the two series may move away from the historical average, and a shift in the inflation target might result in a change in interest rates unless there were good reasons for discounting the excess drift.

Interest rates are thought to be set in relation to the deviation of inflation from its target and also in relation to the presumed gap between actual and potential output. In addition the MPC makes judgements about the risks to inflation that come from factors other than current inflation and the output gap. These judgements, for instance on the potential impact of exchange rate changes, cannot easily be captured by a simple rule. We can estimate a simple rule to capture the roles of inflation, the output gap and judgements in setting interest rates. We can then use this simple estimated Taylor rule to produce a path for interest rates had the MPC been targeting the HICP, rather than RPIX.

Our simple Taylor rule estimated by OLS is:

(1) [R.sub.i] = [alpha] + [[beta].sub.1]([[pi].sub.i] - [[pi].sup.T]) + [[beta].sub.2][[mu].sub.i] + [[epsilon].sub.i]

where [R.sub.i] is the repo rate set by the Bank of England in month i, [[pi].sub.i] the RPIX inflation rate in month i, [[pi].sup.T] is the inflation rate target, in this case 2.5 per cent for the RPIX, [[mu].sub.i] is the output gap in month i and [[epsilon].sub.i] is the error term. The monthly output gap is estimated using the monthly estimates of GDP interpolated from the quarterly data from National Statistics by the National Institute of Economic and Social Research. The output gap was estimated using an approximate band pass filter (see Massmann et al., 2003). It is consistent with our quarterly estimate of the output gap discussed in the UK chapter of this review. (1) The Taylor rule has been estimated since the first decision by the MPC in June 1997-June 2002. (2) The estimated Taylor rule, with bias-corrected t-statistics in parentheses, is:

(2) [R.sub.i] = 5.983 + 1.751 ([[pi].sub.i] - [[pi].sup.T]) + 1.857 [[mu].sub.i] + [[epsilon].sub.i] (17.082) (1.971) (2.052)

The residuals from the Taylor rule were found to be homoskedastic and normally distributed. However, the errors display serial correlation, as we might expect as the judgements of the MPC are likely to be serially correlated. For example, the same argument for interest rates to remain unchanged has often been repeated over a number of months. Despite this, OLS gives unbiased estimates of the parameters. The estimated Taylor rule in (2) lends more weight to the output gap as a predictor of the current repo rate than the difference of the current inflation rate from its target. This could suggest that the MPC see the output gap as a predictor of future inflation, although a simple Granger test of causality does not confirm this.

We use this estimated model to examine what interest rates could have been if the MPC had been targeting the HICP measure of inflation instead of the RPIX. To do this we replace the RPIX variable and its target in (2) with the HICP measure of inflation and an equivalent target for HICP. Although a target of 2 per cent is widely expected, and can be defended as the long-run equivalent of 2.5 per cent for RPIX, we know that the bias over the last six years has averaged 1.0 percentage point, and the MPC has been aware of this. Hence they would perhaps have used a target of 1.5 percentage points, basing the difference on statistical judgements about the information in the indicators. In constructing our estimate of interest rates from the new inflation measure we include the estimated residual as a monthly measure of the judgement of the MPC. This simple method for examining what interest rates could have been ignores the feedback from the change in interest rates to inflation and the output gap.

Chart 3 shows the estimates of what the repo rate might have been over the period June 1997-June 2002 as compared to the actual repo rate set by the Bank of England over this period. There is some tentative evidence that switching the targetted measure of inflation from RPIX to HICP might have introduced a looser monetary policy in 2000 but not at other times.


More obviously, the implications of switching the targetted measure of inflation and strictly enforcing a target close to the long-run equivalent of 2.5 per cent for RPIX (just below 2 per cent), would have been a brief period of lower interest rates, but over the period as a whole would have been associated with higher nominal interest rates.

The volatility of RPIX and HICP inflation

The bands around the target rate should clearly be set in relation to the volatility of the index. If the HICP is a more volatile index than the RPIX then the bands of 1 percentage point as currently used for the inflation target would imply a more conservative approach to monetary policy. Conversely, if HICP is much less volatile than RPIX, then bands of 1 percentage point would imply a more flexible monetary policy.

Chart 4 shows GARCH estimates of volatility for the two inflation measures over time, where periods of high volatility are clearly clustered for both measures in the early 1990s. This period of volatility is also accompanied by wide swings in volatility itself. Chart 4 suggests that HICP is more volatile than the RPIX. Alternatively, this can be shown through simple standard deviations of the two inflation measures for the full sample period. The standard deviation of RPIX over the full sample period is 1.93. In comparison, the standard deviation of HICP over the same period is 2.16. It is plausible that using the more volatile inflation measure, HICP, would mean the 1 percentage point bands currently used for the RPIX measure would be more restrictive for monetary policy, as the greater volatility of HICP may provide the MPC with less room for manoeuvre with interest rate setting. Chart 4 shows the inflation measures stabilised in their level of volatility after the large swings in volatility of the early 1990s. Despite the stabilisation in the level of volatility, HICP still remained the more volatile of the two measures, perhaps unsurprisingly as price stability was defined in terms of the RPIX over this period. The difference in volatility now appears to be negligible. The standard deviations of the two measure since June 1997 are 0.37 for RPIX and 0.38 for HICP, suggesting that adopting the current bandwidth for the HICP would have little implication for the way in which monetary policy is conducted.



We are grateful to Joe Byrne, James Mitchell and Martin Weale for helpful comments and discussion.


(1) Estimates of the output gap depend on the length of the time series, with more recent time periods being revised as new data becomes available. As Massmann et al. (2003) discuss, these revisions can be reduced through extending the time series with forecasts and backcasts. In producing the output gap estimates illustrated on p. 36 of this Review we have been able to use the latest OK forecast. However, with the monthly GDP data we are not able to utilise this resource.

(2) After that date the monthly output gap estimate diverges from the quarterly estimate.


Bank of England (2003) Inflation Report, May.

HM Treasury (2003), UK Membership of the single currency: An assessment of the five economic tests, June, Cm 5576.

Massmann, M., Mitchell, J. and Weale, M. (2003), 'Business cycles and turning points: a survey of statistical techniques', National Institute Economic Review, 183, pp. 90-106.

O'Donoghue (1998), 'Harmonised Index of Consumer Prices: historical estimates', Economic Trends, 541, December.

O'Donoghue, J. and Wilkie, C. (1998), 'Harmonised Indices of Consumer Prices', Economic Trends, 532, March.
COPYRIGHT 2003 National Institute of Economic and Social Research
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2003 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Barrell, Ray; Kirby, Simon; Riley, Rebecca
Publication:National Institute Economic Review
Geographic Code:4EUUK
Date:Jul 1, 2003
Previous Article:The UK economy.
Next Article:Benchmarks and targets under the SGP: evaluating safe deficit targets using NiGEM.

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters