6401.0 - Consumer Price Index, Australia, Mar 2008 Quality Declaration 
ARCHIVED ISSUE Released at 11:30 AM (CANBERRA TIME) 23/04/2008   
   Page tools: Print Print Page Print all pages in this productPrint All




The Australian Bureau of Statistics (ABS) notes the public inquiry into grocery prices being conducted by the ACCC. Considering both the Issues paper released by the ACCC (on 11 February 2008) and a number of the submissions lodged to date, the ABS has prepared the following article to assist understanding of how the Consumer Price Index (CPI) is constructed and alert users to the extensive information available from the ABS website. The ABS is also providing a more comprehensive submission to the ACCC describing some of the relevant key price measures available to shed light on grocery prices in Australia. The ABS submission will be available from the ACCC website <http://www.accc.gov.au>.


The Australian CPI is designed to provide a general measure of price inflation for the household sector as a whole. In practice, the index is constrained to only measure the changes in prices faced by private households living in the six State capital cities plus Canberra and Darwin.

The measurement of price change for an individual, specific item (e.g. Granny Smith apples purchased from a particular store) is a relatively straightforward exercise. It would generally be agreed that an estimate of the average price per kilogram in each period would be all that is required. Price change between any two periods would simply be calculated by direct reference to the respective average prices.

However, problems arise whenever the objective is to provide a measure that covers a number of items (e.g. all apples; fruit; food etc.) or items that are not available in all time periods.

The CPI provides a convenient and consistent way of presenting price information that overcomes problems associated with averaging across diverse items. The overall (or All groups) CPI provides a measure of the average rate of price change, recognising that some items are more important than others.

The simplest way of thinking about the CPI is to imagine a basket of goods and services comprising items bought by Australian households. Now imagine the basket is purchased each quarter. As prices change from one quarter to the next, so too will the total price of the basket. The CPI is simply a measure of the changes in the price of this fixed basket as the prices of items in it change.

The CPI basket covers the full range of household expenditure and is divided into 11 major groups, each representing a specific set of commodities:

    • Food
    • Alcohol and tobacco
    • Clothing and footwear
    • Housing
    • Household contents and services
    • Health
    • Transportation
    • Communication
    • Recreation
    • Education
    • Financial and insurance services.

Cost of Living

Although the CPI is also commonly referred to as a measure of changes in purchasing power or a cost-of-living index, in an economic context these terms are not strictly interchangeable with a measure of price inflation.

An index designed to measure changes in the purchasing power of household incomes would need to be concerned with changes in the costs of all expenditures made from household income. Such a measure would include items like income tax and interest payments. A true cost-of-living index (COLI), among other things, would need to be concerned with the change in the minimum cost of maintaining a given level of utility, or satisfaction, that results from changes in the prices of the goods and services consumed.

A COLI recognizes that the quantities consumed are actually dependent on the prices. In practice, rational consumers may be expected to adjust the relative quantities they consume in response to changes in relative prices (for instance, buying chicken rather than beef when beef prices are high).

The Fixed Basket Approach

The reality is that the actual expenditures of individual households do change over time in response to many factors including changes in prices (particularly relative prices); changes in incomes; changes in family composition (including ages of family members); changes in tastes and preferences and the availability of products (including the emergence of new items). Further, within a calendar year, expenditures can change from month to month due to seasonality of either supply (seasonal availability of produce) or demand (e.g. chocolate at Easter). Changes over time in the expenditures of households in aggregate are additionally affected by changes in the demographic profile of the population at large (e.g. increases in the proportion of older Australians).

At a theoretical level, there are two key approaches that could be used to compile a price index:
  • a fixed utility approach - a basket of goods and services capable of providing the same utility or satisfaction (where the basket is allowed to change but in such a way as to preserve utility or satisfaction); and
  • a fixed quantity approach - a fixed basket of goods and services.

In constructing consumer price indexes national statistical offices, including the ABS, use the fixed basket approach which does not allow the quantities of the individual items to change when measuring period to period price change. The data on the current quantities required for the calculation of 'fixed utility' index are not observable in practice. The 'fixed utility' index is a theoretical index, not an operational one that national statistical offices can produce in a timely manner.

In the case of the Australian CPI, the fixed basket methodology involves devising a basket of goods and services representative of those acquired by metropolitan private households during the course of a full year. The annual basket used in the CPI is based primarily on data obtained from the Household Expenditure Survey (HES) which is the only authoritative source of data on the expenditures of different household types in each of the capital cities.

The composition of the basket and the relative importance of items in it relate to households as a whole - it represents the expenditures of all in-scope households, not the expenditure pattern of an 'average household' or of any particular household type or size.

The HES is conducted at approximately five yearly intervals with the most recent being in respect of 2003-04. Due to the time required to process and validate the data from each HES, the new baskets can only be introduced to the CPI with some delay. The basket based on the 2003 -04 HES was introduced in September quarter 2005 (to measure price change from June quarter 2005). The introduction of new baskets (or item weights) is done in such a way that the index reflects only pure price change and not differences in the cost of the old and new baskets. The 16th series CPI (updated 'fixed basket') will be introduced in the September quarter 2011 CPI based on the 2009/10 HES.

The benefit of the 'fixed basket' approach is that it results in a measure that compares like with like. The downside of this approach is that because it does not allow for item substitutions that consumers are able to make while maintaining utility, it will overstate the rate of price growth compared to a 'fixed utility' approach. The amount by which a 'fixed basket' approach exceeds the 'fixed utility' approach is likely to get steadily larger the further back in time the 'fixed basket' quantities were fixed.

Collecting Prices for the CPI

The collection of prices in each capital city is largely carried out by trained field staff operating out of the various offices of the ABS, while some prices are collected by special surveys out of the Canberra office. Prices are collected in the kinds of retail outlets and other places where metropolitan households purchase goods and services. This involves collecting prices from many sources such as supermarkets, restaurants, travel agents and schools. Prices are collected via personal visit, telephone or internet as appropriate.

In total, more than 100,000 separate price quotations are collected each quarter. The frequency of price collection by item varies as necessary to obtain reliable price measures. Prices of some items are volatile (i.e. their prices may vary many times each quarter) and for these prices frequent price observations are necessary to estimate a reliable average quarterly price.

Each month prices are collected at regular intervals for goods such as petrol, fresh meat, fruit and vegetables, and women's outerwear. For most other items price volatility is not a problem and prices are collected once a quarter.

The prices used in the CPI are those that any member of the public would have to pay to purchase the specified good or service. Any taxes levied on goods or services (such as the GST) are included in the CPI price. Similarly, prices take account of any subsidy or assistance provided directly by government (e.g. Child Care Benefit, Medicare). Sale prices, discount prices and 'specials' are reflected in the CPI so long as the items concerned are of normal quality (i.e. not damaged or shop-soiled), and are offered for sale in reasonable quantities. Any concessions available to particular groups of the population (such as age pensioners) are also taken into account where significant.

Pure Price Change

In constructing a price index the objective is to ensure that the measure reflects only pure price change, in much the same way as retailers would want to compare performance over time on a 'same store' basis. The challenge is to devise a methodology that ensures comparison of prices on a 'like with like' basis.

The use of a basket that contains all of the goods and services acquired over a full year plays a significant role in ensuring the quarter to quarter price movements are on a like with like basis while maximising the item coverage of the CPI. This can be most readily appreciated by considering how items that are only purchased (or available) at certain times of the year might be treated in a price index.

The simplest option would be to exclude all seasonal items from the index and measure price change only by reference to those items available all year round. Such an approach would exclude many items of clothing (summer and winter items), varieties of fresh fruit and vegetables and holiday travel and accommodation to name but a few examples. Although an index that excludes these seasonal items may still provide a reliable indicator of medium to long term price change, it is unlikely to have the same level of credibility and command the same level of respect as an index that includes them.

Inclusion of seasonal items in the basket increases the item coverage of the basket but complicates the task of measuring price change. The key issue is deciding what should be done when an item is out of season (or otherwise unavailable). To simply calculate an average price that excludes the out of season item in one period and then compare that to an average price in another period that includes the now out of season item would provide an erroneous measure of price change. Imagine that the only change between two periods is that the most expensive item is no longer available, but the prices of all other items remain unchanged. An approach that is based on simply comparing the average prices of items actually purchased (sold) would show a price fall when in fact there is no evidence that any prices have changed. The ABS handles seasonal items in the CPI by imputing their prices when they are not available using a technique referred to as class mean imputation. This involves using the price behaviour of similar (or closely related) items to estimate the prices for the seasonally unavailable items.

Measuring Quality Change

The day to day task of measuring price change for individual items has an even greater bearing on the quality of the CPI. The requirement to take account of any changes in the quality of the items priced, to ensure that the index reflects only pure price change, is particularly challenging. While changes to the overall coverage of the CPI basket are only undertaken every five years or so, the individual items actually available in stores are constantly changing. Identifying changes to item specifications and adjusting observed prices to eliminate quality differences or otherwise excluding the observations from index calculations, accounts for a significant proportion of the cost of compiling the CPI each quarter.

In concept quality embraces all the attributes of an item which consumers would consider before making a purchase. For example in the case of tinned tomato soup it would include the volume or weight of the contents as well as the concentration and flavour.

Some changes are relatively easy to deal with while others prove more difficult if not intractable. A marginal change in say the weight of the can of tomato soup from 440gms to 400gms can be handled relatively easily by computing the quality adjusted price by reference to the price per gram. If the list or observed price is unchanged, the quality adjusted price will record an increase of 440/400 or 10%. Quality changes due to either a change in brand or the ingredients pose more difficult measurement problems for which we generally have no ready solution and are forced to treat the change as if it were a change in sample. Some item categories are particularly prone to a high rate of turnover in the specific brands or varieties available, and we are constantly adjusting our samples, again ensuring sample changes are introduced in such a way that the index reflects only pure price change and not differences in the cost of the old and new samples - note that this can be considered as a guiding principle in calculating the CPI.

The Matched Sample Approach

The application of this principle can be best illustrated by reference to a simple though contrived example of how the calculation of a price index for apples may have changed over time. The example assumes that initially only prices of Red Delicious apples are available, then an expanded set including Pink Lady and Gala, and finally back to only Red Delicious and Gala. For simplicity, it is assumed that in each period the quantities are equal. The example can be taken to relate to either a change in sample initiated by the ABS or to an actual change in the marketplace.

Example - Calculation of a price index for Apples

Time period 0
Time period 1
Time period 2
Time period 3
Time period 4


Red Delicious
Pink Lady
Simple average
Matched averages
Periods 0 and 1
Periods 1 and 2
Periods 2 and 3
Periods 3 and 4

Period to period % changes, using

Simple averages
Matched averages

Price index, using

Simple averages
Matched averages

The table above illustrates two possible methods for calculating price change. The first method simply makes use of the average of all prices available in each period while the second method measures period to period price change, based only on the prices of those specific items available in both periods. The second method, which is generally referred to as a 'matched sample approach', is the one that is used to calculate the CPI. The first method, sometimes referred to as the unit value method, represents an approach better suited to tracking changes in expenditure (or revenue), particularly if the quantities purchased (sold) in each period are known. The deficiency of the first method as a measure of price change is even more starkly illustrated if it is assumed that the prices of the different varieties do not change over time, but they do differ from each other.

It is also worth noting that the principle of matched samples is, in practice, also applied to outlets. In other words, even the average price of Red Delicious apples used to measure period to period price movement is constructed by reference to a matched set of outlets, so changes to samples of outlets (e.g. more or less supermarkets) do not of themselves effect the level of the index.

The CPI Does Not Measure Price Levels

The CPI is not designed to measure price levels; rather its purpose is to measure changes in prices over time. While price levels in country regions often differ from those in metropolitan areas (some higher and others lower), the factors influencing price movements generally tend to be similar. Therefore the CPI can be expected to provide a reasonable indication of the changes in prices in Australia as a whole in the longer term.

Similarly, the CPI cannot be used to compare price levels between capital cities. For example, the fact that the CPI All groups index in the March quarter 2008 for Brisbane (165.6) was higher than in Sydney (161.7) does not indicate that Brisbane was more expensive to live in than Sydney. Rather, it indicates that prices in Brisbane had risen more than in Sydney since 1989-90 (when the index was set to equal 100).

At the end of the day, the CPI is most useful as an indicator of price movements, whether it be for specific items, a particular city, or the economy as a whole. The CPI is not, however, a precise measure of individual household price experiences.

Detail Available

As described earlier, the CPI basket is divided into 11 major groups. These are further divided in turn into 33 subgroups, and the subgroups into 90 expenditure classes. An expenditure class is a grouping of similar items, such as various types of motor vehicles. Each of the 90 expenditure classes (for the weighted average of the eight capital cities) is published in index form in table 7 of the CPI publication (cat. no. 6401.0) each quarter. Each of the 90 expenditure classes for each city is published in index form in table 13 (Excel spreadsheet available on the ABS website).

Diagram: Detail Available

Users may also wish to refer to the following explanatory publications that are available free of charge from the ABS website:
  • A Guide to the Consumer Price Index, 15th Series (cat. no. 6440.0)
  • Australian Consumer Price Index: Concepts, Sources and Methods, 2005 (cat. no. 6461.0)
  • Information Paper: Introduction of the 15th Series Australian Consumer Price Index 2005 (Reissue) (cat. no. 6462.0)
  • Information Paper: Consumer Price Index with Reserve Bank of Australia Consumer Price Measures, Australia, 2007 (cat. no. 6401.0.55.002).
  • Consumer Price Index: Historical Weighting Patterns (1948 to 2005) (cat. no. 6431.0)