New Uses of PPP Data Within the EU Eurostat Agenda Item N° 6
Total Page：16
File Type：pdf, Size：1020Kb
Organisation de Coopération et de Développement Economiques Organisation for Economic Co-operation and Development New uses of PPP data within the EU Eurostat Agenda item n° 6 JOINT WORLD BANK – OECD SEMINAR ON PURCHASING POWER PARITIES Recent Advances in Methods and Applications WASHINGTON D.C. 30 January – 2 February 2001 NEW USES OF PPP DATA WITHIN THE EU John Astin, Head of Price Comparisons Unit, Eurostat. ------------------------------------------------------------------------------------------------------- SUMMARY This short paper describes the increasing uses within the European Commission of price microdata collected as part of its regular PPP work. This has led to the opening up of new sources of funding which are expected to lead to benefits for the basic PPP work as well as for the new uses. Given the precarious state of PPP work at the global level, other organisations may find it helpful to consider whether similar “by-products” may extend the scope of PPP work and its potential sources of funding. ------------------------------------------------------------------------------------------------------- 1. For many years, Eurostat has coordinated price surveys required for producing annual PPPs for its multilateral comparison group, now totalling 31 countries. These surveys cover private consumption goods and services, products used in the equipment goods sector, as well as for construction, rents, and public administration. The private consumption surveys have an additional use within the Commission: the calculation of “correction coefficients”, or adjustment factors used to equalise the purchasing power of salaries paid to EC staff serving in different duty stations around the EU. 2. This second use of price microdata has not in general produced any fundamental problems or conflicts of interest, although occasionally this has happened. For example, the products selected for PPP purposes are supposed to be representative of national consumption patterns, whereas those used for correction coefficient purposes should – in principle – be representative of the patterns of consumption of the population of international officials for whom they are designed. Normally this is taken care of by using a different weighting structure; the products themselves are assumed to be the same. But in a few cases, such as international air fares and telephone calls, special products are used for the correction coefficients to reflect international officials’ greater use of international services. As such cases are few and far between, it has not raised any serious problems for the PPP surveys. 3. More recently there have been several new demands for price microdata from within the European Commission. Price dispersion is a function of many variables, inter alia, the degree of competition amongst producers and retailers or the degree of market integration. The Internal Market Directorate-General (DG) needs to have reliable and timely average prices at the level of individual products to evaluate European market integration by measuring price dispersion of similar products sold throughout the EU and changes over time in price dispersion. Regional price differences are also sought. The Internal Market DG also needs to analyse price differences of products purchased by public authorities in Member States to ensure that the public procurement rules work effectively. The Health and Consumer Protection DG is anxious to obtain evidence by which it can judge if the differences between product prices within the EU are such that they jeopardize consumers' interests – and, of course, to identify the reasons behind these differences. A common feature of the needs of both of these DGs is that they need information at item level (average prices) and that (part of) the product basket for which the prices are collected should remain the same over several years to allow comparison of price developments over time. 4. The European Commission is also interested in judging the price-convergence effects of the introduction of the euro in 1999. The effects are expected to accelerate from 2002 when the euro currency is introduced and the existing national currencies withdrawn. It is expected that the transparency of price differentials brought about by a single currency will have an effect on competition and lead to a reduction in price levels, particularly in border zones. 5. There is also a considerable public interest in international price comparisons associated with the growth in international travel for leisure, study, business and employment purposes. This demand is reflected in the increasing growth of press articles – often misinformed and ill-judged – concerning international price comparisons. 6. Private consultancy firms are also showing an interest in these data. For example, Dresdner Kleinwort Benson Research are now publishing regular analyses of price convergence within the EU. Another firm, Ing-Barings, are doing work in the same field. (See, for example, their latest publications attached to this paper). 7. As far as PPPs themselves are concerned, the Regional Policy DG has requested the availability of regional PPPs in time for the next round of Structural Fund allocations, due to be made in 2006. Eurostat is about to award an external contract to study the implications of this new demand in terms of methodology and data requirements. 8. It is clear, then, that there is a wide and growing range of new uses of both PPPs and their underlying microdata which poses both threats and opportunities for mainstream PPP work. 9. Let us consider first the requests for detailed microdata on consumer prices. The data collected for PPP purposes is in principle suitable for the analysis of price dispersion. But in practice there are some fundamental problems. For PPPs, the primary object is to calculate PPPs at the level of GDP and its main aggregates. The calculation process produces PPPs also at lower levels of aggregation down to Basic Heading level, but the main attention is focused on PPPs at the GDP level and on their quality. This has important consequences. Because prices are collected for some 3500 items and then aggregated at various stages, from the theoretical point of view it 3 is not necessary to aim at maximum robustness at the product level (average prices) or even at the Basic Heading level (BH parities): in the aggregation process errors at these levels may be assumed to cancel out, and the reliability of the aggregate results should be sufficient. Thus for a single product only a limited number of items are priced. No statistical error estimates can be made as to the quality of the results. The unreliability of detailed average prices is also the reason why Eurostat has been reluctant to publish results or to make them available at a more detailed level. 10. The new uses relating to consumer protection and price convergence, on the other hand, require reliable data at the individual product level. Up till now, it has not been possible to provide this information from the existing PPP database for the reasons mentioned above. With PPP development in its present phase, it would be impossible to demand the participating countries to make large-scale increases in the sample size of price quotations in order to provide more reliable average prices at the detailed product level. To do so would jeopardise the core work on PPPs. Thus the threat. But of course to have a larger and more reliable sample is precisely what is required to improve PPP quality in general. Thus the opportunity! Eurostat cannot yet pretend to have an answer to this problem. But the way ahead may perhaps lie in the exploitation of new sources of price data. A pilot project – jointly funded by client DGs - is currently underway with three large market research companies (AC Nielsen, Taylor Nelson Sofres, and GfK) to determine whether or not barcode scanner databases can be used for extracting data on prices for a wide range of consumer products. The first results of this project are expected soon. But in principle it seems certain that this approach offers a technique – perhaps the only technique – for obtaining average product prices based on very large sample sizes. (“Very large” in this context can mean millions rather than tens of prices). The larger coverage and number of observations should make it possible to identify not only differences in price dispersion betwee member states but also within them, and hence to indicate if price dispersion between member states differs significantly from that within member states. 11. It is conceivable that the result of this pilot project will be that this technique would be acceptable (and affordable) to the client DGs in the Commission, but not to Eurostat (e.g. for reasons of cost). This would be a most unfortunate outcome. If the technique is successful for the microdata purposes, then it should be incumbent on Eurostat (in agreement with the member states) to find a way of using the same data sources for mainstream PPP purposes. 12. Let us consider next the requirement for measures of price convergence over time. Other papers are on the agenda at this seminar on this specific topic, and the author is not therefore intending to say much on this issue. But one thing needs to be unequivocally stated: PPPs are based on snapshots at a point (normally a year) in time. They are not designed for time series linking and comparison. Research papers have shown that that they cannot be so used. Nevertheless, they are used to make comparisons over time, and as producers we have a duty to users to provide guidance. In the author’s view, it is not sufficient to tell users that they should not make temporal comparisons of annual PPP results. We should be producing results which optimise the spatial and temporal aspects of PPPs. PPPs are the only way of assessing price-level convergence over time, and there is a clear and growing demand for such 4 data.