6419.0 - Producer and International Trade Price Indexes, 1995
ARCHIVED ISSUE Released at 11:30 AM (CANBERRA TIME) 13/02/1995  Ceased
Page tools: Print All
Contents >> Chapter 17. Other issues

Introduction

17.1. Apart from information on the variety of component series of index numbers, there is a potential demand for two other kinds of statistics relating to prices. These are:

• average prices of individual goods; and
• indexes showing relative price levels between different cities or other localities (sometimes referred to as spatial price indexes).

Average prices

17.2. For a large proportion of the items priced for the producer and foreign trade price indexes, it is not possible to calculate meaningful average prices. The reasons for this are:

• the nature of the sampling procedures; and
• the difficulty of defining a meaningful average price for many items.

17.3. As explained in Chapter 12, the ABS price indexes are designed to measure price change over time. To do this, samples of products are selected to provide representative indicators of price movement over time and not to measure the actual average price at any date. Thus the average price of the sample for instant coffee is simply the average of the prices of a number of the biggest selling brands and sizes, selected to measure price change for instant coffee. This average could, in no way, purport to be the overall average price for all sales of instant coffee. To calculate an actual average price would require prices to be collected for a much larger range of the brands and sizes of instant coffees produced.

17.4. Similarly, the samples of respondents from whom prices are collected are selected to provide representative measures of price change over time and not necessarily to provide accurate measures of the actual average price level for the item concerned. As a general rule, the sample required to provide a reliable measure of price change is smaller than that required to measure accurately price levels at a date. This reflects the fact that, in determining the size of the respective samples, two different sets of assumptions apply:
• in the case of a sample to measure price change over time, the assumption is that the prices of suppliers, etc., not in the sample move similarly (on average) to the prices of those in the sample, or are not significant in terms of value of transactions; or
• in the case of a sample selected to measure price levels, the assumption is that those not in the sample either charge (on average) the same prices as those in the sample, or are not significant in terms of value of transactions.

17.5. In general, the first set of assumptions can be met with a smaller size sample than the second.

17.6. There are also many items where it would not be practicable to calculate an average price for any recognisable thing which could be usefully composed. This may be because:
• the variety of different items purchased and sold is so diverse; and/or
• the methods of pricing are so complex and diverse.

17.7. Example of such items include clothing, appliances, machinery and equipment, furniture and floor coverings.

17.8. Because of these considerations, the ABS does not calculate average prices from the producer and foreign trade price indexes. A limited number of average prices is available from the Consumer Price Index.

Spatial price indexes

17.9. The term spatial price indexes is used to describe indexes which measure the difference in prices between localities (i.e. differences across space) at a particular date - in contrast with the more common temporal price indexes which measure differences in prices over time in a particular locality. Although both the Materials Used in House Building and the Materials Used in Building Other than House Building indexes provide indexes for each capital city, they are temporal and not spatial. That is, they provide price movements over time for each city, not price movements between the cities.

17.10. Compilation of spatial price indexes is more difficult than is generally realised and encounters both theoretical and practical problems. No generally accepted statistical methodology has yet been developed for compiling indexes of this kind. In theory, satisfactory spatial price indexes could be compiled if average price levels could be calculated accurately for an identical range of goods and services in all of the centres to be compared, and if a satisfactory basis could be found for combining these prices (i.e. for weighting). However, in practice these conditions are not likely to be realised.

17.11. Where two localities have different patterns of sales or purchases (which is the common situation) it is not possible to determine an appropriate single regimen for measuring prices or an appropriate single weighting pattern for combining the prices in each of the two localities. Differences in sales/purchases patterns do not involve merely different proportions of sales/purchases of different items. It commonly happens that some items important in the sales/purchase pattern of some localities do not exist at all in others.

17.12. In attempting to construct an index of relative prices between two localities ('A' and 'B') the regimen and weighting pattern could conceivably be based on:

• the sales/purchases pattern in locality A (i.e. the basket of goods purchased in A would be priced in both localities and the relative costs compared); or
• the sales/purchases pattern in locality B (i.e. the basket of goods purchased in locality B would be priced in both localities and their relative costs compared); or
• some kind of average of the regimen and sales/purchases patterns of the two localities.

17.13. In general, neither alternative (1) nor (2) could be regarded as providing a satisfactory basis (especially from the point of view of the locality whose sales/purchasing pattern was not taken into account). Alternative (3) would be regarded, at best, as a compromise which reduces, but does not eliminate, the shortcomings of (1) and (2). Extending the index to more than two localities compounds the difficulties.

17.14. Even if the problem of determining a regimen and weighting pattern for a set of spatial price indexes were to be resolved, there would still be a substantial practical problem in actually constructing such indexes because identical types of goods and services are not necessarily sold in each locality (i.e. the brands, styles, sizes, etc., differ from locality to locality). While this is not a problem in compiling temporal price indexes designed to measure price changes in each locality separately, it is a serious difficulty if the objective is to compare price levels between localities. Unless identical goods and services could be priced in each locality it would be necessary to assess the extent to which price differences reflect differences in quality or quantity and adjust prices accordingly before using them in the comparison.

17.15. Because of the difficulties mentioned above, the ABS has limited its involvement in spatial price indexes. The only such indexes the ABS produces are indexes of relative retail prices of food; generally the problems of obtaining prices for identical items in different localities are less serious in the case of food than in other areas of expenditure.

You have reached the end of publication 6419.0

 Previous Page