CPI measures and retail trade changes

MESI Feature Article

Consumer prices in the Monthly Statistical Bulletin

The Monthly statistical bulletin has traditionally reported the Australian Bureau of Statistics (ABS) measure of consumer prices—the Consumer Price Index (CPI). Starting with the October 2009 edition, Table 2.4 will be expanded to include a measure of ‘underlying’ or ‘core’ inflation, based on statistical manipulations of the CPI designed by the Reserve Bank of Australia (RBA).

Consumer prices generally

The CPI measures the relative cost of a basket of goods and services purchased by capital city households. The basket includes up to 100 000 goods and services allocated to 11 categories, consisting of 90 expenditure classes. These classes are weighted in terms of their relative importance in households’ budgets. The basket is updated every five years to incorporate changes in the composition of household purchases.

Changes in the CPI measure inflation in Australia, and this measure is often referred to as ‘headline inflation’.

Reserve Bank measures

Headline inflation incorporates price classes which change by large magnitudes for reasons other than a shift in value.  For example, fruit is seasonal and also can be affected by extreme weather events, both of which affect availability and therefore the price paid by consumers.

To address this issue, the RBA created measures which smooth these volatile fluctuations to try and estimate a ‘core’ or ‘underlying’ rate of inflation. These measures are the ‘trimmed mean’ and ‘weighed median’.

Basically, the trimmed mean is calculated in stages.  First, the percentage changes in all CPI classes are ordered by their relative size.  Then, the top and bottom 15 per cent of classes, by percentage price movement, are excluded from consideration. Finally, the remaining middle 70 per cent of classes are used to create a weighted average, or ‘mean’.

The weighted median is effectively a secondary measure.  The middle 70 per cent of classes from the trimmed mean process is the benchmark set of data.  The weighted median measure is the price change observation that is in the middle of this benchmark set of data. 

In public reporting it is typical that the changes in these two measures are averaged to provide the estimate of core inflation. This convention has been adopted in this bulletin.

The underlying or core rates of inflation are important as they are, typically, the rates considered by the Board of the RBA when they pursue an ‘average two to three per cent’ inflation rate target, using monetary policy instruments like the overnight cash interest rate.

An example of observed differences

Graph 1 compares quarterly changes in the average of the RBA measures and the CPI.  The key point is that the RBA average tends to be smoother than the CPI changes and, therefore, is much more like a trend rate of change.

Graph 1: Quarterly change in prices—RBA and ABS

Graph 1: Quarterly change in prices—RBA and ABS

Monthly statistical bulletin Table 2.4

The bulletin print edition Table 2.4 has been updated to include index values, quarterly and annual changes for the CPI, as well as the average of the annual underlying rates of inflation. The print edition graph will show the annual underlying and headline rates of inflation including a visual identification of the 2-3 per cent RBA inflation target range.  In addition, the electronic database will report both of the RBA measures in more detail. 

Changes to retail trade

The ABS implemented changes to the scope of the retail trade survey in its July Retail Trade publication.

The changes resulted from adopting a revised set of Australian New Zealand Standard Industrial Classifications (ANZSIC). ANZSIC 2006 changed the definitions of some industries compared to its predecessor, ANZSIC 1993. In addition, some activities which were retailing were transferred into alternative classifications and some activities which were not previously retailing were bought into scope.

The data in the Retail Trade publication are reported on the new ANZSIC categories. The effect is a consistent upward shift in monthly retail trade.  The changes were backcast to April 1982. Further changes have been foreshadowed, including publication of trend estimates and a change in seasonal weighting patterns. When these changes occur further revisions to this bulletin may also be required.

Table 4.1, in the print and electronic editions, has been updated to include retail trade data based on the revised ANZSIC basis. This means that data in Table 4.1 of this edition of the bulletin should not be compared to earlier print or electronic editions, as the data are inconsistent.

Adrian Makeham-Kirchner

Statistics and Mapping Section

Back to top