Part 1: A technical note prepared by the Specialist Adviser

1. This note demonstrates that the HICP will always be lower than the RPI, and that the difference becomes larger, the greater is the difference in the inflation rates of the individual items that make up the index (i.e. the variance of the individual item inflation rates). For example, a shock that affects the inflation rates of the individual items differently, such as a change to the exchange rate, will make the two indices diverge.


2. RPIX is an arithmetic weighted average of price relatives. HICP is a geometric weighted average of price relatives.


3. The weights are previous expenditure shares. In the RPI these are up-dated annually, and are not the current expenditure shares. Thus the weight on the ith good is:

8. This can be interpreted as:

9. If the rates of growth of the individual prices (the individual inflation rates) are not identical, the RPI will be greater than the HICP.

or as

10. the greater the differences in the rates of growth of the individual prices, the greater the divergence between RPI and HICP.

11. In practice, the individual price changes will tend to be different when either demand or supply conditions are different. An example is an exchange rate change. This is likely to directly affect the traded goods and services sectors, but not the non-tradeables sectors. Any effect on the non-traded sectors will be indirect and hence later.

12. Although there may be differences between the RPI and the HICP, if financial markets understand the differences, switching from using RPI to HICP ought not to have much effect on interest rates.

The Retail Prices Index (RPI)

The Retail Prices Index is the main domestic measure of inflation in the United Kingdom. It measures the average change from month to month in the prices of goods and services purchased by most households in the United Kingdom. The expenditure of certain higher income households and of pensioner households mainly dependent on state pensions is excluded, as their expenditure patterns are considered atypical.

The three major uses of the RPI are as follows:

• A macroeconomic indicator to monitor inflation.

• Income adjustment such as uprating social security benefits, state pensions, and the capital value of index linked National Savings and gilts, and for wage bargaining.
• Price adjustment, for example setting prices in private contracts and regulating prices of privatised utilities.

The index is compiled using a large and representative selection of more than 600 separate goods and services for which price movements are regularly measured in 146 areas throughout the country. Some 120,000 separate price quotations are used each month in compiling the index.

The RPI is constructed using two forms of arithmetic mean. The average of relatives is used when prices have a wide range or goods are dissimilar, whilst the ratio of averages is used when prices are similar or goods are homogeneous. The methodology underlying the construction of the RPI has been developed and approved by RPI Advisory Committees over many years.

Once published the RPI is never revised.

RPI excluding mortgage interest payments (RPIX)

RPIX is the measure used as the government's inflation target. It is constructed in exactly the same way as the RPI except that it excludes mortgage interest payments. It is a more appropriate tool for monetary policy purposes because by excluding mortgage interest payments it does not reflect the direct impact of interest rate changes made to control inflation.

The harmonised index of consumer prices (HICP)

The harmonised indices of consumer prices (HICPs) are calculated in each member state of the European Union for the purposes of European comparisons as required by the Maastricht Treaty. The official series starts in January 1996. From January 1999 it has been used by the European Central Bank to measure price stability across the Euro area. Within the United Kingdom, its main use is for comparisons of inflation with other EU countries.

The construction of the HICP is specified in a series of EU Regulations agreed between Eurostat (the Statistical Office of the European Community), the European Central Bank and Member States, which are intended to ensure that the indices from different countries are comparable. The comparability of methods is designed to ensure that the calculation of annual inflation rates does not differ for methodological reasons by more than 0.1 percentage points.

The HICP uses the same basic data as the RPI but with some important differences in coverage and methodology:

• the geometric mean is used to aggregate prices;
• It excludes a number of RPI series most notably those relating to owner-occupiers housing costs (mortgage interest payments, house depreciation, council tax and buildings insurance). The HICP also currently excludes some health, education and insurance expenditure - these will be introduced in December 1999;
• the HICP includes air-fares which are currently excluded from the RPI;
• The HICP index for new cars is a quality-adjusted index based on published prices of new cars, whereas the RPI uses used cars as a proxy for prices of new cars. The latter was recommended by a previous RPI Advisory Committee;
• Weights for insurance, particularly the service charge element, are constructed according to different rules. For instance in the HICP the weight for claims expenditure is distributed according to type of expenditure (eg car repairs, spare parts) whereas in the RPI it is presented under specific insurance heading (eg house contents, motor);
• The average household expenditure pattern on which the HICP is based is that of all private households (and therefore includes certain higher income and pensioner households, which are not covered in the RPI).

Provision is made in the Commission Regulations for the HICP to be revised. This enables retrospective adjustment to the coverage of the index

Which index for which purpose?

The ONS recognise that no single measure on inflation can meet all users' needs. That is why we publish a variety of measures. The RPI is the "headline" measure of consumer price inflation as it measures the widest range of price changes that affect the majority of private households. It is the main domestic measure of inflation in the United Kingdom. RPI(X), which excludes the impact of mortgage interest payments, is arguably a better indication of "underlying" inflation for monetary policy purposes and is used by the Government for its inflation target. The HICP is calculated in each Member State of the European Union according to standard methodology and is therefore the most appropriate for making comparisons within the EU since such comparisons are not hindered by differences in the way each country compiles its index.

Choice of aggregation formula

The choice between geometric and arithmetic means for combining price data at an elementary (unweighted) level depends both on the purpose for which an index is intended and its conceptual basis. This requires a judgement about the relative statistical merits of the formulae concerned. An index constructed using one formula is not necessarily more accurate than an index constructed using a different formula. In the development of the HICP no judgement was made on the technical merits of each formula: the United Kingdom HICP uses the geometric mean because it produces comparable results between countries and helps to ensure that differences in inflation rates are not due to differences in the formulae used to create indices at the most basic level. One of the two types of arithmetic mean used in the RPI (the ratio of averages) also has this property and can also be used in the HICP. The other type of arithmetic mean (the average of relatives) leads to non-comparability and is therefore barred.

In February 1999, the different formulae used in the HICP and RPI accounted for about half of the 0.9 percentage points difference between HICP and RPIX annual inflation rates (the derivative of the RPI most closely matching the HICP in coverage). The remainder of the difference is due to coverage (see below).

		Difference between annual rates HICP-RPIX		Breakdown of differences (unrounded figures)
		rounded figures	unrounded figures	housing components excluded from HICP	insurance weights (net vs gross)	formula effect1	other coverage differences
1998	Dec	-1.1	-1.08	-0.40	-0.13	-0.56	0.01
1999	Jan	-1.0	-1.03	-0.41	-0.10	-0.52	0.00
	Feb	-0.9	-1.00	-0.44	-0.09	-0.51	0.04

¹ Difference due to use of different formulae to aggregate prices at the most basic level.

The three inflation measures described above are price indices which are constructed from raw price data and which are designed specifically to measure price changes. The GDP deflator and consumers' expenditure deflator - commonly referred to as "implied" measures - are not directly calculated from raw data. They are derived from a comparison of National Accounts economic series measured in current and constant (eg 1995) prices.