Changes in the seasonality of Retail Turnover

The ABS publishes seasonally adjusted statistics so that underlying monthly and quarterly movements, and non-seasonal influences, are easier to see in our statistics.

Seasonally adjusting data means estimating and removing seasonal and calendar related effects from original data. For retail, this means removing the usual impact on turnover of events like Christmas or Easter and it helps us to compare turnover from one month to the next. Other examples of seasonal and calendar effects include seasonal changes in weather and the number of days in a month. 

The Black Friday sales event is another example of a regular seasonal event. It is also an example of how new regular events can cause seasonality to change over time. 

The seasonality of retail turnover has changed over time as spending and trading patterns changed. For example, shopping on Sundays has become more popular than it was 20 years ago as trading laws have changed, and more stores are open. This means that for many retail categories turnover is much higher than it used to be in months when there are five Sundays. 

Graph 1 below shows that spending patterns for non-food related retail turnover in the months around Christmas were relatively stable for much of the past 20 years. The onset of the COVID-19 pandemic in 2020, however, caused an abrupt change in spending patterns that led to extreme movements in turnover. 

These spending changes were temporary and were mostly related to responses to public health actions, such as lockdowns. There were large changes in turnover for some retail businesses when lockdowns either came into effect or ended. 

Temporary and irregular changes in spending, like COVID-19 effects, are included in seasonally adjusted data. This means that in the seasonally adjusted data there are some very large movements which reflect the impact that irregular events of the pandemic had on turnover as shown in graph 2 below.

Accurately identifying and measuring the impact of these temporary changes on the seasonality of retail turnover has been difficult. At the same time as spending patterns were distorted because of COVID-19, Black Friday sales events were growing in popularity. This change not only affects the sales pattern in November, but also in the surrounding months. 

Black Friday sales started in the United States and became prominent in Australia a bit over ten years ago for a small number of online businesses as a one-day sales event on the fourth Friday in November. From around 2017 more mainstream online businesses started to become involved, as well as some bricks and mortar stores. It is now a large sales event for both online and bricks and mortar stores, and not just for one day, but for an extended period. In 2023, some retailers were branding their sales events as Black November. 

The growing popularity of Black Friday sales in recent years means that its influence on retail turnover and seasonal patterns has been increasing over time, although this influence varies across different retail categories. This has caused seasonal factors, which are created during the seasonal adjustment process, to gradually change to reflect the evolving seasonal pattern as shown by graph 3 below. 

To seasonally adjust data, the ABS uses a method known as concurrent seasonal adjustment. This method uses the most recent data available to measure seasonality and create seasonal factors. We apply seasonal factors to original data to get seasonally adjusted data. Please see retail methodology for more information on seasonal adjustment. 

We can’t know what is a seasonal versus irregular impact on turnover using one year of data only. For the most recent retail data we use this year’s data and the previous three-years to estimate seasonality and calculate seasonal factors. 

Where spending patterns are changing rapidly, this will be reflected in changing seasonal factors, however, because we need to use a few years to measure the seasonality, seasonal factors will gradually change to reflect shifts in seasonality over time. Where seasonal patterns change abruptly and quickly settle at a new level, it can be appropriate to apply an adjustment known as a seasonal break. A seasonal break was added in 2019 for some retail categories to reflect the initial impact of Black Friday sales on seasonal patterns from 2017. 

Our view for the more recent data is that although there is evidence to suggest seasonal patterns have changed significantly over the past two years as turnover in November has jumped to a new level, there is not enough stable data to reliably measure a seasonal break. Instead, allowing the seasonal factors to evolve gradually reflects the ongoing growth in the popularity of Black Friday sales and clearly shows changes in spending patterns.

This is reflected in the results for this year’s Black Friday sales. There has again been strong growth in the seasonally adjusted data published for November 2023 and the seasonal factors have grown further. This shows that the popularity of this event has grown again this year. At some point, it is likely the growth in the popularity of Black Friday sales will level-off. The seasonal factors would then update to account for this new level of sales as a regular seasonal effect and it would not have the same influence on seasonally adjusted data. 

An associated impact of the growth in popularity of Black Friday sales is that the months around November are also experiencing a change in seasonal pattern. The results vary depending on the retail category, but using clothing retailing as an example in graph 4 below, shows turnover has: 

• Become weaker in October as consumers hold off on some spending ahead of the discounts available in November.
• Remained relatively steady in December. 
• Become weaker in January as the post-Christmas sales are not attracting spending at the same level as they were previously. 

As spending patterns continue to evolve, if users want to look at underlying activity it can be better to look at quarterly seasonally adjusted data or monthly trend data. This is because monthly seasonally adjusted data is impacted by changing seasonal patterns as described above, as well as one-off events that might have impacted only one month of turnover. 

Because seasonal factors evolve over time and are calculated using several years of data, as more information becomes available there will be revisions to seasonal adjusted data. This reflects improvements in the data as the seasonal pattern becomes clearer and distinct from irregular events. Given the large changes in spending patterns around Black Friday, revisions for November and the surrounding months in the coming years are likely to be higher than usual. 

We will continue to regularly review our treatment of Black Friday sales to ensure our estimate of seasonally adjusted data provides the best indication of the underlying story in retail trade.