Revision of the Price Index of First Hand Domestic Sales (pif)
The price index of first hand domestic sales is going through a major revision and alteration these days. This revision covered most of the areas within the index, which means that the users also were affected.
The revision was effective from the January index, published 9th February 2001.
Improvement for the user
The revised index will imply more improvements for the users. The methods of computation are altered in several areas. Geometric mean is introduced in the micro level estimation, for technology products the hedonic method will be used for quality adjustments (inter alia computers), a chained Laspeyres formula is employed for aggregating and so on. An annual updating of the weights will be introduced, and the standard used will be updated.
A price index of first hand domestic sales will be published the way it is done today, that is a price index indicating the price development in the Norwegian market and from imports as a whole. The revised index will in addition publish indices indicating the domestic price development.
The revision yielded a break in the series
The price index was, due to the alteration, based on certain different principles and methods. The transition to a new index was evaluated as so important that it is accurate to refer to the alteration as a break in the series. However, earlier indices with 1977=100 and 1981=100 were chained to the new index so that the historical development is intact.
Due to the break in the index one had to calculate changes across this breakpoint in two steps. Firstly, the change from the selected month and year to December 2000 was calculated. Secondly, the change from the recalculated December 2000 to the chosen month and year after the break was computed and multiplied by the former.
The official index including December 2000 with 1981=100 will be available at our web site.
A brief examination of changes due to the revision
In the following a brief examination of the main areas in the revision is carried out. The areas are:
- Method of calculation
The index adopted the latest version of the SITC - Standard International Trade Classification - the UN classification for statistical purposes. In the SITC the commodities are grouped by level of processing.
An important change was that the indices according to use ceased. However, there is instead a decomposition of the producer price index into four main industrial groupings - intermediate goods, capital goods, consumer goods and energy goods. The consumer goods are in addition divided into durable consumer goods and non-durable. This may be an alternative for certain purposes.
The foundation of weighting is, as before, the National Accounts' numbers for production and imports.
The revision introduced a yearly updating of the weights. In the calculations of detailed levels the weights are based on the final National Accounts' figures. Based on the quarterly National Accounts' data the weights are projected until the year in advance of the index year.
The change to a yearly updating of weights means that the sectors that normally have large variation in prices and/or the value of production from one year to another may have larger variation in the contribution to the index calculations.
Methods of calculation
Geometric average: At micro levels we employ a geometric average. This is in accordance with international recommendations. Studies of methods used in calculation at micro levels have proven that a geometric mean is suitable for price statistics. The method is also evaluated as more robust when changes are monitored in the primary material.
Hedonic method: When buying e.g. IT products many purchasers observe that the nominal prices are relatively stable over time. At the same time they see that the product quality is considerably developed. In the price index of first hand domestic sales and other price statistics it is the quality adjusted price development we want to measure. For this we have in the work with the index, in addition to other methods, adopted the hedonic method. The basis for this method is that the technical characteristics or other qualitative aspects of the product are closely related to the price. Using regression techniques based on price information and technical data allows for separating the price changes from the quality changes.
Firstly we will use this method for calculation of the price development of computers and washing machines. Subsequently, we will adopt this method to other technology products with large quality changes.
Data from Foreign Trade: The Foreign Trade Statistics have a considerable material (value, volume etc.) associated with exports and imports of commodities. For some of these the Foreign Trade price material is evaluated as a suitable indicator for measuring price developments.
In the future, price material from the Foreign Trade and other suitable price information will be considered and employed where relevant. In some areas price information gathered directly for price statistics purposes (e.g. some spot prices) is used.
Chaining: The index uses, as before, a Laspeyres-formula when aggregating above micro levels. The traditional formula with fixed weights and products was replaced by a chained index. This has to be viewed in context with the yearly updating of weights.
The change to a chained Laspeyres formula is connected to the fact that a traditional formula over time will complicate the updating of the product sample. This work will be considerably easier with a chained solution where changes are simply made in December in connection with the yearly revolving of the production system. This contributes to a faster introduction of new products. A chained solution combined with a yearly updating of weights makes the measurement of the price development more correct, especially over time.