Experimental statistics Early indicator of short-term economic trends based on pedestrian frequencies

Examining the relationship between official retail turnover and pedestrian figures

8 August 2022 - The coronavirus pandemic has hit the German economy hard, especially shops in inner cities. Since mid-March 2020, many retail shops were either closed for some time or were open only under strict conditions and for a limited number of customers. At the same time, the pandemic-related economic slump and the resulting massive short-time work led to a sharp decline in the consumption expenditure of the population. Due to this development, turnover in shop-based retail trade slumped several times (see Figure 1). In the same period, the numbers of pedestrians in shopping streets often fell considerably due to the pandemic and the political measures taken in this context. This is shown by daily data received by the Federal Statistical Office from the hystreet.com GmbH company for selected metering points in shopping streets (see Figure 1). These data are published regularly in the "Dashboard Deutschland".

Figure 1

^{Note on the figure 1: The rates of change in pedestrian frequencies include very high values for winter/spring 2021/2022 (base effect). This can be explained by the fact that various lockdown measures which were in force in Germany between November 2020 and May 2021 resulted in very low pedestrian figures in the examined inner cities.}

©Grafik: © www.dashboard-deutschland.de / Datenquelle: © hystreet.com GmbH

The data from hystreet.com can be used to show the development of the numbers of pedestrians in German inner cities. In many German cities, the company counts pedestrians by means of laser scanners operating 24 hours a day. The system counts people with a height of over 80 centimetres. Destatis currently receives data from five metering points in the five largest cities of Germany (Berlin, Hamburg, Munich, Cologne and Frankfurt on the Main). In the analyses shown here, Frankfurt is not included as data are missing there for a longer period in spring 2021. However, all data available from that metering point are shown in the Dashboard Deutschland.

The development of shop-based retail turnover can be shown by means of official monthly data based on a sample of all enterprises whose main activity is trade (see table "Turnover figures in shop-based retail trade as of 2018"). The figures become available 30 to 45 days after the end of the reference month and cover a wide range of retail sectors and branches (see Section G of the Klassifikation der Wirtschaftszweige 2008 (WZ2008)). The benefit of official data is that they are collected in a standardised process in accordance with legal provisions and that they are then checked for plausibility and edited. This means that they are representative and meet high quality standards in terms of accuracy.

The decreasing turnover observed in shop-based retail trade and the slump in pedestrian figures in inner cities raise the questions of whether there is a statistical relationship between the two figures and of whether pedestrian figures can thus be used as a very early short-term indicator for this branch of retail trade as they are based on near real-time measurement. This would be the case if people went to inner cities and their pedestrian zones primarily for shopping in retail stores. Falling pedestrian figures would then lead to decreasing turnover in shop-based retail trade. Vice versa, retail trade would probably benefit from highly frequented inner cities with many pedestrians. In its experimental statistics (EXSTAT) section, the Federal Statistical Office will continuously monitor this connection between numbers of pedestrians and shop-based retail trade. This is to examine whether the data can be helpful for official statistics beyond the pandemic in the medium to long term, that is, to serve as an early economic indicator showing economic trends at the current end better and earlier. Also, based on the development of pedestrian figures, it might be possible to draw conclusions regarding customers switching to e-commerce. The greatest benefit of such a very early indicator would lie in the fact that pedestrian figures are available on a daily (or even hourly) basis and thus earlier than the official monthly survey of retail turnover and many other early indicators.

Figure 1 illustrates that the development of shop-based retail turnover correlates with the development of pedestrian figures in German shopping streets. The year-on-year percentage change rates of turnover show slumps in the months after the lockdown measures came into effect, including shop closures. Before the pandemic, retail trade usually recorded rising turnover year on year. It becomes clear that changes in the numbers of pedestrians in shopping streets generally coincide with retail shop openings and closures.

The decrease in pedestrian figures is markedly larger than the decrease in retail turnover. One of the reasons is that, although people still go to convenience stores (such as supermarkets), the latter usually are not located in inner cities, which means that their customers were not counted by hystreet.com. It is also possible that a number of retailers whose shops were closed due to coronavirus-related measures switched to e-commerce at short notice, which was not yet reflected by the data collected. In addition, the data from retail trade statistics and the pedestrian figures are not seasonally and calendar adjusted; consequently, part of the two time series might be affected by seasonal and calendar effects. Only the relevant shopping times from Monday to Saturday between 10 a.m. and 8 p.m. have been taken into account for pedestrian figures.

Figure 2

The statistical relationship between the two change rates is rather strong, with a correlation coefficient (Pearson’s R) of roughly 0.7 (see Figure 2). However, with a few exceptions, decreases in turnover were recorded only in 2020 and 2021 – both were coronavirus years with restrictions imposed on retail trade. The development of pedestrian figures was very similar. This means that currently the correlation between the two time series is heavily influenced by the coronavirus pandemic circumstances, involving shop closures, decreasing consumption expenditure and declining mobility of the people. The measures of dispersion also show the extent of the impact the coronavirus crisis had on the individual development of the two time series. In the pre-crisis year of 2019, the standard deviation of the variables was relatively small, whereas dispersion has been growing since February 2020, as is illustrated in Table 1.

Figure 1 also shows that the short-term economic trends in the whole of Germany can be reflected rather well by these four metering points, even though shopping streets are not included in the whole of Germany. The correlation is quite similar in all cities, with a slight variance of the correlation coefficients (see Table 2). It should be noted, however, that turnover data are available only at the federal level, so that they are identical for each city in the evaluation - variance is shown only for the numbers of pedestrians. The underlying assumption is that the development of turnover is roughly the same in all shopping streets and that none of the cities markedly deviates from the national development of turnover.

The correlation between the two time series confirms that changes in pedestrian figures are connected with changes in retail turnover. No causal relationship can however derived from that fact. A number of adjustments and analyses will be required before a reliable, very early indicator can be produced for shop-based retail trade. The fact that there is a clear correlation became obvious only through the pandemic-related restrictions (shop closures, slump in consumption expenditure). Examining the rather small number of data points before the pandemic (May 2019 to January 2020) shows a much lower correlation. This is also due to the generally low variance of both time series in that period, as is shown in Figure 3.

Figure 3

More information about the correlation between pedestrian frequencies and retail turnover and about whether this is suitable as an early indicator might be obtained by examining the correlation with other official data on retail turnover. As an alternative from official statistics, "Turnover of retail trade with other goods in stores" would be an obvious choice. It does not include supermarkets and gives a better picture of the shops in the branches that were affected most heavily by the coronavirus-related restrictions. Figure 4 shows that the correlation there really is higher so that further studies may be promising.

Figure 4

^{Note on the figure 4: The rates of change in pedestrian frequencies include very high values for winter/spring 2021/2022 (base effect). This can be explained by the fact that various lockdown measures which were in force in Germany between November 2020 and May 2021 resulted in very low pedestrian figures in the examined inner cities.}

Overall, however, it is not possible yet to give a final assessment of the correlation observed. The time series are still too short and too much affected by the coronavirus crisis. This is why the Federal Statistical Office will continue to observe the development of the correlation. Also, the time series cannot be adjusted yet for seasonal and calendar effects as the time series of pedestrian figures is short. It is planned to adjust the time series as soon as sufficient data points are available. It will then become clear whether pedestrian figures can be a reliable short-term indicator of the development of shop-based retail turnover.