On 11 October 2021, a 'hybrid' meeting, Measuring EU Inflation – The Foundations of the HICP, was attended by people at the RSS in Errol Street and remote viewers. It was organised by the History of Statistics and Official Statistics sections of the RSS.
RSS fellow John Astin, previously head of the Prices Division at Eurostat (1991-2002) gave the talk. He outlined the contents of his recently published book with the same title. The meeting was chaired by Tony Cox, Chair of the Statistics User Forum's RSS RPI/CPI User Group.
The Treaty of Maastricht (1992) specified that a harmonised measure of inflation should be developed and that by 1997 all member states should publish inflation information regularly according to this measure; a working party was set up to carry out the development. It consisted of members of John’s Eurostat team and delegates from EU and EFTA member states. Various task forces were set up to deal with specific issues. In nine years, all these bodies together met nearly 100 times! The HICP was not to be an index of household spending, but an inflation measure for international comparisons and a macroeconomic indicator. Most of the member states were very protective of their existing systems for measuring inflation and chose to keep on using them alongside the HICP.
John started with a brief history of attempts, going back to the ILO in 1925, to get international measures of the cost of living produced.
Among the more important technical issues is product classification. The working party used a slightly amended version of a new international classification developed by the UN and OECD. Also, coverage was an issue: the price paid by the consumer in the country (not overseas) was what was to be recorded. A few legally dubious items were excluded. As far as possible, the whole population would be included.
Collecting prices was a major difficulty. One problem was 'missing prices' for products no longer available was a problem; so were seasonal products. Quality adjustment raised serious difficulties where two products are functionally similar but (say) are made of different materials. Finally, in 2007 a regulation dealt with this issue.
There are many products for which, when specified in great detail, no weights are available and so the returns can’t be aggregated. The staff have to calculate an artificial index for the class of products using different types of average – in most cases only Jevons or Dutot averages are allowed.
The treatment of owner-occupied housing is a long-standing problem. The majority of states used imputed rents. The final recommendation was to use net expenditure on houses, but this still isn’t in the HICP.
The weights to be used are obtained from regular household budget surveys. There are two alternative ways of creating a national average from these: household-weighted ('democratic') or expenditure-weighted ('plutocratic'). The HICP uses the latter, in keeping with its purpose as a general macroeconomic indicator.
The ONS uses weights based on the year before last: John is concerned about what will happen next year, using weights from 2020 when strange things happened to expenditures.
John went on to review the expansion of the EU: he added a brief survey of HICP developments after he left Eurostat.
He finished by describing the Household Costs Index, a new experimental index which the ONS is developing. It is intended as a 'democratic' index like the RPI and can be disaggregated into different household types. It has a number of important differences from a macroeconomic index like the HICP. John suggested that everyone should look at the experimental indices published so far by ONS.
Paul Smith, Professor of Statistics at Southampton University, was lead discussant. From his comments and John’s replies came a few points:
- The accuracy of the published figures couldn’t justify more than 1 decimal point
- The choice of formulae needed a majority of at least 62%; it received exactly that in qualified majority voting
- HICP is revisable
- Disagreements could be referred to the EU Council, as happened in 3 or 4 cases
- The domestic expenditure concept avoids double counting
- The book is a good read, even in the more esoteric areas - a helpful tour of challenging topics in consumer prices.
John’s responses to comments from the audience included:
- If he had his time over again, he would he do things the same way
- UK Treasury is pushing to have the RPI replaced by the HCI because the latter would give lower inflation figures
- He wouldn’t comment on a possible change to the system for dealing with seasonal goods, as he wasn’t an expert; he thought net acquisitions as a measure of housing expenditure would come in eventually.