Issues concerning the use and measurement of the HICP
Prof. Eugenio Domingo Solans, Member of the Governing Council and of the Executive Board of the European Central Bank (ECB), Speech delivered at the CEPR/ECB Workshop on Issues in the Measurement of Price Indices, Frankfurt, 16 November 2001
Ladies and gentlemen,
The pleasure I have in addressing and spending some time with you in this workshop on measurement issues concerning the Harmonised Index of Consumer Prices (HICP) is twofold. First, because it concerns statistics, a subject in which I have a keen interest. Second, taking into account the present circumstances, because it has nothing to do with euro banknotes and cash changeover issues. Maybe I should rather say that "hopefully" HICP developments have or will have nothing – or at least not much – to do with the euro cash changeover.
I am aware that some of you have travelled a long way to be present here today: a special welcome to you. I would also like to thank all those who have contributed papers or are acting as discussants. In particular, I would like to thank those who have been most active in organising the workshop – here I should mention, in particular ERNST BERNDT, (Massachussets Institute of Technology) and DIETMAR HARHOFF (Universität München) from the Centre for Economic Policy Research (CEPR) side, and GONZALO CAMBA-MENDEZ of the European Central Bank (ECB), and, of course, VITOR GASPAR, who first had the idea for such a workshop and has been closely involved at every stage. Needless to say, I am personally very grateful to PETER BULL who, in addition to his extensive involvement in other aspects of this workshop, was even – would you believe it? – involved with the speech I am giving you now.
You have had a long day of discussions and I do not intend to keep you from your dinner for too long. Nevertheless, the HICP is of such importance to the ECB that I should like to make a few remarks about it. I hope that my remarks – also made from the perspective of a member of the ECB's Executive Board – will complement those made by Otmar Issing in opening the workshop.
Any workshop on measurement issues is likely to elicit much comment about the quality of the statistics concerned, their shortcomings, and ideas for improvement. Indeed that is the purpose of a meeting of this kind. As a main user of price statistics, the ECB wishes to monitor whether there are criticisms of the data of which it was not aware, and naturally it wants to hear ideas for improvement, especially ideas of the rare variety that can be implemented quickly and at no cost!
I would like to touch upon the three main functions which, inter alia, the HICP performs: as a measure for assessing the nominal convergence of a country wishing to qualify for membership of European Monetary Union, as a quantitative definition of the prime objective of the ECB's monetary policy – by far the most important HICP function from the ECB's perspective – and, finally, the use of the HICP as a benchmark for indexing economic and financial magnitudes.
The assessment of convergence was the reason for developing the HICP, and in some of its features it betrays its origin. Assessing convergence required, above all, comparability across European Union (EU) Member States: in reaching the all-important decision concerning readiness to proceed to Monetary Union, it was vital that the analysis should not be confused by arguments about whether the rate of inflation in country A had the same meaning as the same rate measured in country B. In order to avoid this kind of dispute, Eurostat and national price statisticians elaborated a rather detailed framework, supported by legislation, to regulate the collection and compilation of price data. As far as possible, they limited any scope for discretion which might have given rise to dispute. We know that there is unfinished business – like quality adjustment – but they were wholly successful in providing unchallenged numbers for the convergence exercise in spring 1998.
I should perhaps note that this function – the assessment of convergence – will, as far as we can see, continue, because three current Member States have not yet adopted the euro and more countries are likely to join the EU and become potential participants in the euro area in the coming years. The potential future enlargement of the euro area and the need to assess – as far as possible – aspiring entrants on a consistent basis over time, is in practice a reason for caution in changing the HICP. I should like to emphasise that we very much welcome the progress made in introducing the HICP in the accession countries; this will help the ECB and the Commission to carry out the tasks involved in assessing convergence and its degree of sustainability as established in Article 109 j(1) of the Treaty on European Union and even more clearly in Article 1 of Protocol No. 6 to the Treaty. The HICP will help the accession countries to compare themselves with the EU and the euro area. Indeed, the HICP is now produced by more than 25 European countries, easing cross-country comparisons across this large group of countries with close economic and financial relations.
The HICP – despite the measurement issues which are being discussed in this gathering – was the natural choice for the ECB's definition of price stability. Given the ECB's main responsibility under the Treaty on European Union, namely to maintain price stability, it is not surprising that we have taken a close interest in the development of the HICP, practically ever since the European Monetary Institute (EMI) was set up in 1994. Indeed, the HICP provides a good example of the successful implementation in practice of the principles of co-operation to be found in various pieces of Community legislation, including the 1995 HICP Regulation itself, which requires the ECB to be consulted on harmonisation issues.
At this point, I should like to draw attention to the close co-operative relationship which exists between Eurostat – in charge of the elaboration of the HICP – and the ECB. Eurostat and the ECB are the two European institutions responsible for providing statistics at the EU level and we highly value our relationship.
A price index covering the euro area as a whole was what the ECB needed at the start of its work. The HICP had other attractive qualities. It covered a large proportion of household expenditure; it was available monthly and in a timely manner; and it was, or soon became, aggregable in another sense, namely that the country pieces fitted together without gaps or overlaps. It was also subject to either no or only minor revisions. The ECB liked two other features of the HICP – the absence of imputation (it reflects "monetary transactions") and the fact that it is a price index, not a deflator, although, of course, with frequent adjustment of the weights. (In mentioning these last considerations, I realise that the ECB's preferences were not uncontroversial.) The HICP was the only serious contender for the measurement of inflation in 1998, given the ECB's need for aggregability, monthly frequency, timeliness and reliability.
Allow me to make a few further brief remarks on the ECB's use of the HICP. The ECB regards its obligation to maintain price stability as a medium-term one – we do not, of course, have any means of doing so on a month-by-month basis. Moreover, we know that monetary policy operates with those famous long and variable lags. Essentially, we must try to predict the behaviour of the HICP some way ahead – taking account of all the influences that bear upon it – and take policy action if we foresee persistent departures from price stability.
The customary way to express inflation is the increase in the index over the past 12-month period; and this is the measure to which we have also given prominence. This measure is easy to understand and best known; and, of course, it minimises the effect of short-term seasonal influences. The disadvantage is that it is as much affected by influences on the index a year ago as it is now; by definition, it may not tell us whether prices are currently accelerating or decelerating. There is a certain awkwardness in explaining monetary policy changes in the present, intended to influence inflation one to two years ahead, in terms of the behaviour of the price index for the preceding month as compared with twelve months earlier.
It was partly to address this difficulty that the ECB began to publish staff macroeconomic projections at the end of last year. Around the same time – in the January 2001 edition of the ECB's Monthly Bulletin – we published, for the first time, seasonally adjusted HICP data, elaborated by ourselves; these constitute a useful tool to better understand short-term changes. We are cautious in using the seasonally adjusted HICP series and continue to give prominence to the 12-month comparison in published comment; however, we can now analyse movements in prices over shorter periods with some confidence. As an analytical tool – but certainly not as an alternative benchmark for our definition of stability – I personally find the seasonally adjusted HICP percentage change on the previous period – month or quarter – which we publish in our Monthly Bulletin very useful. I find it useful, despite being fully aware of its limitations. It gives me an idea of the present or marginal "speed" of inflation, which complements the information on the average "speed" since the "journey" started one year ago. It may be as useful, but also as misleading, as using the information shown on the speedometer of my car at any given moment to find out when I will reach my destination.
As I mentioned earlier, the objective is to maintain price stability, defined on the basis of the increases registered by the HICP, over the medium term. This reflects the consensus that monetary policy cannot fine-tune developments in prices over short horizons and, therefore, that the ECB's monetary policy will not react mechanically to a specific set of monthly HICP data.
Nevertheless, monetary policy decisions cannot be isolated from market expectations and economic agents' psychology. A specific set of monthly HICP data, especially if it is a surprise, can affect expectations and must therefore be taken into account by policy-makers. In other words, in spite of the medium-term orientation of monetary policy, a single set of HICP data matters as do, of course, its accuracy and timeliness. In this respect, I welcome the publication by Eurostat of a "flash indicator" of the HICP, which began with the October 2001 data. The publication of "flash indicators" is a correct way to address the issue of striking the right balance between accuracy and timeliness, one of the many choices good statistical frameworks are confronted with.
In short, because of its original importance in assessing convergence, the HICP is far more closely harmonised than other commonly used measures of prices and costs, and much more readily used to support the conduct of monetary policy than national account deflators. These are the reasons why the HICP plays a key role for the ECB and has been used to define price stability, and it is of course a key indicator for the second pillar of the ECB's policy strategy. Again, good reasons to improve it, if necessary, but also a reason for caution in changing it.
Now, some five years after the HICP came into existence, other developments can be observed. The use of the HICP is widening, and we can observe additional non-monetary policy users and uses. In particular, its function as a benchmark for indexing economic and financial values and variables is being developed. One example is the recent announcement by the French government that its new bond is to be indexed to the HICP for the euro area as a whole. Their choice surely reflects the view that the HICP is now established as the prime index in the euro area – with national CPIs playing a less relevant role – and that the use of the euro area aggregate, rather than the national HICP, is likely to widen the bond's investor appeal.
Let me be clear in this respect. The national measures of inflation or deflation play a prominent role as an informative and administrative statistical tool and should continue to do so in the future. I am not opposed to such measures; on the contrary, I recognise that the national measures of price developments fulfil a necessary function. What is increasingly hard to justify, in my view, is the existing dichotomy between comparable and harmonised HICPs at the national level and non-harmonised national CPIs. The simple existence of two indexes for the same purpose seems odd and confusing in itself. The elaboration, on the basis of similar criteria (mutatis mutandi), of a comparable European HICP and national HICP for each country would certainly be better and cheaper. This would in all likelihood facilitate the increasing administrative use of the European HICP instead of the national ones. Little by little, euro area economic agents will realise and accept that their natural economic and financial habitat is the euro area as a whole and not their own country, and will tend to move to the European HICP to index economic and financial magnitudes, in line with the French example I mentioned before.
Let me now draw a comparison between the HICPs and other macroeconomic statistics for the euro area. Considering the tremendous amount of work that has been carried out over the last eight years to produce good quality HICPs, it is no surprise that HICPs are rarely mentioned in the widespread critical comments on euro area statistics. The HICP was not included in the Action Plan on statistics presented to the EU ministers of finance, the purpose of which is to achieve early improvements in the areas most needing attention. I believe that, by contrast with some other areas of economic statistics, the European HICP compares well with the US consumer price index. However, I am glad to see that discussion of the latest progress report in the ECOFIN Council a few days ago strongly supported the work which Eurostat has already undertaken to improve quality adjustment in the HICP. Quality adjustment methods are essential for the comparability of the HICP. For some product groups, such as electronic equipment, clothing and cars, it has to be assumed that the differences are very significant. It is likely that the problem has increased as a result of the speed of technical innovation. Work on the behaviour of some sub-indices across the EU shows differences in measured price changes which can plausibly be explained only by a different allowance for quality changes. It must be considered, however, that (a) solving the problem is extremely difficult and often resource-intensive, that (b) overall, the positive and negative quality bias may cancel each other out (e.g. cars and housing - where it is probable that the quality of the rented stock deteriorates with age) and (c) the problem still affects only a certain and limited part of the index, because in other areas the change in quality is slow or quality adjustment easier. The ECB fully supports Eurostat's increased efforts to make progress in this field, and also the work to measure the cost of owner-occupied housing. Thus, a number of challenging tasks lie ahead of the statisticians, and it is our hope that this seminar can give some impetus to this work.
In conclusion, it is clear that the HICP has played, is playing and will indeed continue to play a tremendous role in our economies and societies. Important values, such as price stability, important developments, such as European monetary integration, and important administrative uses, such as indexation, will eventually resort to the HICP. The lesson to be drawn from the importance of this statistical tool is that its uses are such that a degree of caution must be exercised in developing it further. This does not, of course, mean that the HICP should not improve and evolve – failure to do so would in time make it unsuitable for any purpose. However, the HICP is too important to be changed without careful reflection.