3. PRODUCER PRICES
3.8 Problem areas
There may be many reasons why PPI samples are unrepresentative and thus liable to provide inaccurate results. All national PPIs suffer from collection and compilation problems to a greater or lesser extent. Problems associated with missing prices and seasonal items have been dealt with in Part 2.12 and there is no need to repeat them here. Examples of other problems are:
• samples are selected purposively rather than using probability sampling methods, increasing the chances of bias. For example, establishments may be selected for their convenient geographical location or because they are known to be good respondents;
• without probability selection methods, estimates of statistical accuracy cannot be made (but without some initial estimate of variance, a randomly selected sample cannot be optimised either so that stratification must be based on judgement);
• the sample size for an industry may have become outdated if the industry has grown or contracted since the base period, i.e. the period when the sample was selected;
• new products may not be identified or included in the survey. This problem may be relieved to some extent by rotating the sample of establishments;
• the sampling frame may be out of date or may not include certain groups of the target population. A common problem is that information on small producers is unreliable as this group is often volatile, with the result that the weight for small producers may be wrong. Typically they are under-represented.
In an ideal world, it would always be possible to use statistically sound sampling techniques to produce PPIs of the required accuracy, within given resource constraints. Reality, however, usually does not conform to this ideal. It is usually impossible to correctly optimise samples since reliable estimates of population variances are rarely available, sampling frames are always deficient to some extent, and response rates are unpredictable.
The aim is, therefore, to make the best use of what is available and to apply the principles of sampling in a common-sense and practical way. Arguably the most important step in sampling is to fully establish and understand what the survey is trying to estimate, the limitations of the sampling frame and the environment in which the survey will be conducted, i.e. likely response rates, data quality, and the level of available resources.
Once this starting position has been established a sample design can be drawn up, with decisions being made about stratification, sample size and allocation. Random sampling techniques may be employed in countries where a large amount of data are available and reasonable estimates of variance can be made but it is usually the case that samples are selected judgementally. If done wisely, this may be a perfectly reasonable approach.
As with most panel samples, PPIs suffer from the problems associated with a changing population. Any sample of establishments and products will become increasingly unrepresentative over time, and is likely to be depleted as establishments cease production. Some form of panel rotation or supplementation is advised to minimise any bias caused by these problems.
Changes in quality
Ideally, the index should not be affected either by changes in quality or changes in the sale conditions. Although it is not always possible to achieve this objective, the following procedures are useful in trying to separate pure price movements from other changes in a situation where a particular product is replaced by a substitute product of different quality. In all cases, the judgement of the national statistical institute officer and his knowledge of the particular product are of vital importance:
• if the two products have been available for some time on the same market, both having sold in reasonable quantities and having fairly stable prices, it can be assumed that the price difference between the products is attributed to a quality change. The new series is simply spliced to the old one;
• if the two products are not available at the same time or if their prices have been unstable, the ratio of production costs of the two may be used along with judgement based on this and other information provided by the manufacturer to separate the change in the price from the change in quality.
As with the compilation of consumer price indices, the diversity of techniques applied by any one country in treating quality changes for the various goods priced for PPIs prevents direct comparison of each methodology actually used. The following analysis of the methodologies is, therefore, restricted to determining whether or not any adjustments are undertaken and to outlining some of the main methodologies used.
Table 20: Producer prices: Methodology for treatment of quality changes
Summary of main methodologies used
Canada ..
Mexico Differences between products are identified to deduct from the new price the part that correspond to the additional characteristics and unit prices are obtained for those products that change in weight or volume.
United States Production cost method1and hedonic regression method2for technology products
Australia Mainly production cost method
Japan Mainly production cost method, hedonic regression method for technology products from 1990 base index
Korea Mainly production cost method, hedonic regression method for technology products
New Zealand No adjustments are made
Austria Adjustment of the base prices
Belgium No adjustments are made
Czech Republic Overlapping pricing3, linking
Denmark The price index for a new commodity is chained to the index for the Old commodity
Finland ..
France ..
Germany Direct valuation of differences in quality characteristics, overlapping pricing, linking.
Greece ..
Hungary No adjustments are made
Iceland ..
Ireland ..
Italy A correction factor is applied based upon prices surveyed for the new and old varieties in a month of overlap.
Luxembourg ..
Netherlands Overlap imputation.
Norway Using a new base price (December price of the previous year) or estimating using the price development of the old product, or in the group. Hedonic methods for computers.
Poland No adjustments are made
Portugal No adjustments are made
Slovak Republic ..
Spain ..
Sweden Using a new base price (December price of the previous year) or estimating using the price development of the old product, or in the group. Hedonic methods for computers.
Switzerland No precise information about method.
Turkey No adjustments are made
United Kingdom Overlap method