ESA/STAT/AC.94/Bkg-2 18 November 2003 UNITED NATIONS DEPARTMENT OF ECONOMIC AND SOCIAL AFFAIRS STATISTICS DIVISION Meeting of the Expert Group on International Economic and Social Classifications New York, 8-10 December 2003 Requirements for product classifications used when making observations of household expenditures and when collecting price observations for and calculating a consumer price index, with possible implications for the Central Product Classification Eivind Hoffmann, ILO Bureau of Statistics
2 International Labour Office, Bureau of Statistics Bureau International du Travail, Bureau de statistique Oficina Internacional del Trabajo, Oficina de estadística http://www.ilo.org/stat http://laborsta.ilo.org stat@ilo.org fax: + 41 22 799 6957 Requirements for product classifications used when making observations of household expenditures and when collecting price observations for and calculating a consumer price index, with possible implications for the Central Product Classification 1 Introduction When developing the Central Product Classification (CPC) the main statistical applications that have influenced the work seems to have been statistics on production and on international trade. Other areas of statistics where it is important to have an appropriate classification of products (i.e. goods and services) include statistics on the expenditures of households, collected through household expenditure surveys (HES) and other sources, as well as the Consumer Price Index (CPI). The purpose of this note is to present briefly the type of requirements that these areas of statistics would like a classification of products to satisfy. The general methodological starting point for this note is that a classification is a discrete set of values that can be assigned to one or more variables or characteristics that can be observed for statistical units. For some of these value sets it is convenient to create a classification structure that will describe how a large set of discrete values (categories) can be grouped together to more aggregate sets or groups. 2 The more explicitly formulated the criteria used for aggregating the most detailed categories to a groups at higher levels in the classification system, the easier it is to understand where new units should be placed (classified) once their characteristics have been observed. This is particularly important for the CPI and HES as new items, both new variants of existing items and genuinely new goods and services, are emerging more or less continuously in many economies. It should be possible to handle such new items as soon as they appear, and it should be possible to include observations of their prices in the compilation of the CPI as soon as the total expenditures on them become significant. 1 Note prepared by Eivind Hoffmann as a contribution to discussions at the United Nations Workshop on Classifications, Santiago, Chile, 8 12 September 2003. It expands on text originally prepared by Peter Hill and Fenella Maitland-Smith for the revised manual on CPI being prepared by a group of experts and an Inter- Secretariat Working Group on Price Statistics (forthcoming). The attached figure describing a typical CPI aggregation structure was prepared by Valentina Stoevska on the basis of figure 3 (p. 30) in Turvey, R. et al (1989). The opinions expressed are those of the author and do not necessarily reflect those of the ILO Bureau of Statistics. Comments and suggestions for improvements will be welcome. E-mail: Hoffmann@ilo.org 2 Note that (1) by defining similarity criteria for aggregation to higher level groups we also define criteria for distinguishing between them; and (2) the criteria used to distinguish categories at the aggregate levels in the classification system must be present also in the distinctions made at the most detailed level, or as descriptors attached to the most detailed group. Relevant descriptors are needed for alternative aggregation structures.
3 Classification requirements for the CPI and HES When collecting prices for the CPI each price observation has to be described in terms of several characteristics or variables. In addition to the price itself the most important of these characteristics will be the type of item, the amount to which the price applies, the type and location of the outlet, and other conditions of sale than the price. Of these variables the type of item and the type and location of outlets are the most relevant ones for the construction of weights to be used for compiling the CPI, as requests are often made for sub-indices defined in terms of one or more of these variables. The classification system for type of item will provide a structure that is essential for many stages of CPI compilation. Most obviously, it provides the weighting and aggregation structure for the index. In addition, it provides a scheme for stratification of products for price observations, whether by random or non-random sampling, down to a certain level of detail; and it dictates the range of sub-indices available for publication. There are several factors which must be taken into account when a CPI classification system for type of item is being developed. The classification system must be able to reflect economic reality. For example, it must be possible to accommodate new goods and services in a manner that minimize any need for restructuring of the higher-level categories. Many users require long time series, and such restructuring of the classification will produce breaks in the series. This also means that any changes to the classification structure should be introduced together with other changes that will lead to changes in the time series, e.g. when introducing new weights (if this is done at longer intervals only). The needs of users for sub-indices should be given a high priority when constructing aggregate groups: e.g. if some users are particularly interested in price changes for food products, then the classification should make it possible to have a separate index for this area, as long as the weighting data and price observations can support the requested level of detail. The categories should be unambiguously mutually exclusive, and at the same time provide complete coverage of all items within the scope. In practice this means that it should be straightforward task to assign any particular items, type of expenditure, and the associated price, to a single category of the classification system, and preferably at the most detailed level in the classification system. The availability and nature of statistics on expenditure patterns, e.g. as observed in a HES, should be consistent with the classification system for the CPI. The availability of weighting data as well as price observations will dictate the lowest level of detail for which the calculation of a separate index might be possible. Obviously it is not possible to produce a separate product index for a class of products for which the number of observations is not sufficient to estimate expenditure weights with the required degree of precision, or for which a sufficient number of price observations cannot be made.
4 Similarity criteria for classifying expenditures and prices by type of consumption item The compiler of the CPI first collects data about individual products and then aggregates them according to one or more classification schemes, as illustrated by the attached flow diagramme. The first step in this process is to create elementary aggregates of items that are similar with respect to expected price movements, among items that are regarded as similar with respect to the type of needs that they satisfy. Price movements are partly a consequence of (a) changes to the cost of production, e.g. changes to the costs of factors of production such as labour and raw materials as well as changes in the production process itself as a consequence of technological changes; and partly a consequence of (b) changes in demand, e.g. needs. Thus for the elementary aggregates (the lowest level of aggregation relevant for the CPI calculations) the main similarity criteria will be that the items are the same or that they satisfy the same need and are produced with similar raw materials and production processes: e.g. one elementary aggregate may be leather footwear of various types (similarity criteria: type of need and type of raw material), and for the first corresponding higher level aggregate the leather footwear may be combined with footwear produced mainly from other raw materials (similarity criterion: type of need). At the top level of the classification system to be used for the CPI and HES the consensus among CPI experts seems to be that one would want to have needs aggregated according to the different purposes in a manner that can achieve consistency with the System of National Accounts (SNA-93). This means that the preferred aggregate classification structure for purpose will be one that is consistent with that of the International Classification of Individual Consumption According to Purpose (COICOP). In practice there may also be demand for higher-level sub-indices according to the physical properties and intrinsic nature of the items, as well as the main industrial origin of the items. These criteria were used as when creating the structure of the CPC 1.1; and the last is also reflected in the International Standard Industrial Classification of All Economic Activities (ISIC, rev. 3). Classification by purpose according to COICOP COICOP makes distinctions at the highest level by 12 broad categories of purpose; and below this level the groups and classes of the classification are presented as product types. The implied assumption therefore is that each product type can be assigned to one and only one type of purpose among the 12. However, in practice prices (and expenditures) are observed for multi-purpose products (single products that can be used for a variety of purposes), such as electricity; as well as for bundled products, such as package holidays comprising transport, accommodation, meals, etc. Thus rules were formulated for the handling of such products. Multi-purpose goods and services In drawing up COICOP, the general rule followed was to assign multi-purpose goods and services to the division that was considered to represent the predominant purpose.
5 Hence, motor fuel is shown under Transport in COICOP. Where the predominant purpose varies between countries, multi-purpose items was assigned to the division that represents the main purpose in the countries where the item concerned is particularly important. As a result, snowmobiles and bicycles are both assigned to Transport because this is their usual function in the regions where most of these devices are purchased - that is, North America and the Nordic countries in the case of snowmobiles; and Africa, South East Asia, China and the low-countries of Northern Europe in the case of bicycles. Examples of other multi-purpose items in COICOP include: food consumed outside the home are classified under Hotels and restaurants not Food and non-alcoholic beverages ; camper vans, which are shown under Recreation and culture, not Transport or Housing ; and basket-ball shoes and other sports footwear suitable for everyday or leisure wear, which are shown under Clothing and footwear, not Recreation and culture. The COICOP guidelines encourage national statistical offices to adapt to national circumstances the manner in which products are allocated to the COICOP purpose categories if they consider that an alternative purpose is more appropriate in their countries. Such reclassifications should be footnoted. Bundles of goods and services Single outlays may sometimes comprise a bundle of goods and services which serve two or more different purposes. For example, the above mentioned food consumed outside the home seems to be an example both of multi-purpose (nutrition and leisure) and of bundling (food & service). Other examples mentioned are: the purchase of an all-inclusive package tour, which will include payments for transport, accommodation and catering services; the purchase of educational services, which may include payments for health care, transport, accommodation, board, educational materials, etc, in addition to the actual teaching activities; the purchase of in-patient hospital services which include payments for medical treatment, accommodation and catering; and the purchase of transport services which may include meals and accommodation in the ticket price. In all these cases COICOP recommends that such items should be dealt with on a case-by-case basis with the aim of obtaining an overall purpose breakdown that is as precise as possible and consistent with practical considerations of data availability. Hence, purchases for package holidays are shown under Package holidays with no attempt to isolate separate purposes such as transport, accommodation and catering. At its most detailed (class) level COICOP is making a distinction between services (S), non-durables (ND), semi-durables (SD) and durables (D). This supplementary classification is intended for analytic applications, and may be relevant also for HES results and CPI sub-indices. Implications for work with CPC
6 The main implication of the above is that if CPC is to provide a framework for the international comparison of statistics dealing with goods, services and assets and to serve as a guide for developing and revising existing classification schemes of products in order to make them compatible with international standards, see paragraph 10, section IIA of (UN, 1998), then further work will have to be done to examine to what extent and how the CPC sub-classes can serve as elements for the elementary aggregates for the CPI as well for the different types of higher level aggregates mentioned above. The work already carried out by Statistics Austria and OECD in mapping CPC categories to COICOP will be one useful starting point, although HES and CPI applications may not have been explicitly considered in that work (OECD, 2001). One issue that needs to be examined is whether it will be possible and useful to introduce (additional) distinctions in CPC to explicitly facilitate the separation of products that (potentially) fall within the scope of HES and CPI from (similar) products that clearly are outside this scope, e.g. because they are only used as inputs to a production process. Another implication is that guidance on how to up-date national product classifications that are to serve HES and CPI applications will be needed, both in general and also on how to best ensure consistency with and support for the work that the UNSD Classification Section (hopefully) will be undertaken to up-date CPC on an regular basis. References: OECD (2001): COICOP CPC and CPC COICOP: Correspondence tables. OECD, Paris. Turvey, R. et al (1989): Consumer price indices: An ILO manual. International Labour Office, Geneva. UN (1998): Central Product Classification (CPC). Statistical Papers, Series M No. 77, Ver 1.0. United Nationas, New York. Geneva, 29 October 2002 (revised 2 July 2003)
Example of typical CPI aggregation structure All items CPI (100%) 7 Major group index Food (23%) Major group index Beverages (10%) Major group index Other major consumption groups (67%) Higher level index Bread and cereals (14%) Higher level index Meat (3%) Higher level index Others (6%) Item group Rice (5 %) Item group Bread (2%) Item group Other cereals (7%) Regional Rice sold in northern region (3%) Regional Rice sold in southern region (1%) Regional Rice sold in other regions (1%) Elementary aggregate Rice sold in Supermarkets in northern region (2%) Elementary aggregate Rice sold in independent outlets in northern region (0,6%) Elementary aggregate Rice sold in other outlets in northern region (0,4%) Representative item Parboiled white rice Representative item Brown rice Representative item Black rice Variety brand A, model X Variety brand A, model X Variety Others