Enhancing household consumption measures in the National Accounts



Following an internal review of Household Final Consumption Expenditure (HFCE) the ABS received Government funding to enhance the quality of National Accounts HFCE estimates with a focus on using more alternative data sources in the compilation of HFCE.

In recent years, there have been rapid advancements in the availability of new data sources. Transaction level data sets are becoming increasingly common in Australia and around the world. These datasets offer enormous benefits for National Statistical Offices (NSOs) due to their high frequency, coverage and granularity. These datasets have the potential to reduce collection costs and the reporting burden placed on households and businesses. The use of these datasets in the compilation of HFCE offers significant advantages over existing sources and can enhance the quality of HFCE estimates.

Current HFCE data sources

Currently HFCE is compiled by using a range of data sources to produce both “benchmarks” and “indicators”. Data sources such as the Household Expenditure Survey (HES) and the Retail and Wholesale Industry Survey (RIS/WIS) are used to compile the benchmarks and set the level of HFCE. These data sources generally have good coverage of household consumption and associated product detail enabling accurate measurement. However, due to the size and cost of these surveys, they are run infrequently. For example, the HES is run every 6 years. As a result, the benchmarks are only infrequently updated.

To produce annual and quarterly estimates of HFCE, higher frequency data sources with less detail and coverage are used as indicators to move forward benchmarks. For example, the monthly Retail Trade Survey provides timely data on household consumption, however, there are limitations in using the Retail Trade Survey as an indicator of household food consumption. The survey collects industry turnover and does not collect any information about the products sold, which reduces the ability to pick up changes in the underlying composition of household consumption.

Figures 1 and 2 provide a summary of the current data sources in terms of benchmarks and indicators and their contribution to the overall compilation of HFCE. These illustrate a reliance on the use of traditional ABS surveys for the majority of HFCE compilation, with administrative data sources accounting for only a small proportion. Due to the lack of timely data, modelling techniques such as trend interpolation have to be used in compiling annual and quarterly indicators (see figure 2).

Figure 1: Data sources used for HFCE benchmarks (a)

The image is a pie chart representing the numerical proportions of data sources used for HFCE benchmarks.
The image is a pie chart representing the numerical proportions of data sources used for HFCE benchmarks. The proportions are as follows: Retail and Wholesale Industry Survey (RIS/WIS): 36% Household Expenditure Survey: 24% Census: 18% Admin data: 13% Financial collections: 9%

a. Based on 2020-21 HFCE current price levels.

Figure 2: Data sources used for HFCE indicators (a)

The image is a pie chart representing the numerical proportions of data sources used for HFCE indicators.
The image is a pie chart representing the numerical proportions of data sources used as HFCE indicators. The proportions are as follows: Retail Trade Survey: 35% Modelled: 32% Admin data: 16% Financial collections: 9% Business Indicators Survey: 8%

a. Based on 2020-21 HFCE current price levels.

Experimental estimates of HFCE food using supermarket scanner data

Since 2011, the ABS has received “scanner data” from major supermarket chains in Australia. This data has been used extensively in the calculation of the Consumer Price Index (CPI) contributing approximately 16% to its compilation. More recently, the ABS has been developing methods and processes to produce experimental estimates of HFCE Food using the supermarket scanner data.

In level terms, the experimental estimates are, on average, 7% above the currently published estimates. This reflects differences in the datasets currently used to compile HFCE (primarily RIS/WIS and the Retail Trade Survey) compared to the scanner data. Detail on the experimental estimates including the conceptual framework and methods can be found in Using scanner data to estimate household consumption, September 2021.

Using the product level detail in the scanner data, changes in food consumption patterns over time can be calculated. Figure 3 compares the experimental and published HFCE food estimates. The increased difference from 2020 onward can be attributed to increased food sales during the COVID-19 period.

Similar differences are evident when comparing growth rates, most notably in March quarter 2020 (figure 4). For HFCE food, the experimental estimates appear to improve accuracy compared to the existing Retail Trade indicator, as food sales are more accurately measured from the mix of products purchased from supermarkets.

Index reference period: September 2016 = 100.0

New analytical insights

In addition to improving measurement, the use of scanner data in the compilation of HFCE estimates provides new insights into consumption by type of product (see figure 5).

The scanner data can also be used to produce insight into “quality’ changes. The term “quality” can be subjective, however, for the first time the ABS has produced a quantitative measure of “quality” allowing for analysis into consumer substitution behaviour. This analysis is described in Recent applications of supermarket scanner data in the National Accounts.

The analysis shows that over time, the quantities of food consumed by households remained relatively stable. Changes in quality drove the changes in overall volumes of consumption. This was particularly apparent from June 2019 to March 2020, when consumers reduced the quality of food purchases (figure 6). Changes in quality tended to be inversely related with changes in food prices. That is, in times of rising prices, consumers tended to reduce the quality of the food consumed.

In June quarter 2020, volumes of food consumption returned to more normal levels following significant stockpiling activity by households in March. Scanner data showed that households substituted to higher quality food purchases (e.g. higher quality ingredients), presumably in response to the closure of dine-in services at cafes and restaurants and more time cooking at home.

Next steps

Experimental estimates of HFCE food using scanner data are published with the September 2021 quarterly National Accounts. These are available for download in the Data downloads section of the release, under Data cubes. The estimates will remain experimental for the next 12 months whilst continuing to assess the accuracy and coherence with other data sources. The ABS plans to implement these estimates into the September quarter 2022 National Accounts to be released in December 2022. These estimates will replace the existing quarterly indicator HFCE food based on the Retail Trade survey.

The scanner data development work is now completed. The ABS is continuing to explore and increase the use of alternative data in the National Accounts. The immediate focus will be on developing methods and processes for incorporating bank transactions data in the compilation of quarterly HFCE, for selected categories. These categories include parts of Recreation and Culture, Operation of Vehicles and Other Goods and Services. Other alternative datasets are also being explored, including transaction data from major retailers. The ABS will update users on the development of this work in the coming months through information papers.

For more information or any queries relating to the experimental estimates and their implementation, please email national.accounts@abs.gov.au.

Back to top of the page