Growth and Cycles
Project: Consumption Indicator
RWI-Konsumindikator zeigt schwächere Entwicklung des Konsums an
Pressemitteilung vom 5. März 2013
Der RWI-Konsumindikator ist im vierten Quartal 2012 von 54,8 auf 40,7 gesunken. Dies deutet auf eine schwächere Ausweitung der privaten Konsumausgaben im ersten Quartal 2013 hin, denn der Indikator weist erfahrungsgemäß einen Vorlauf von einem Quartal gegenüber der Vorquartalsrate des Privaten Konsums auf. Für das zweite Quartal 2013 deutet der Indikator auf eine nochmalige Abschwächung hin; in den ersten beiden Monaten des ersten Quartals ist er auf 38,9 gesunken. Da der Indikator darauf ausgelegt ist, Tempowechsel in der Entwicklung der Konsumausgaben anzuzeigen, lässt die Stärke des Rückgangs allerdings nur begrenzt Rückschlüsse auf das Ausmaß der Abschwächung zu.
Bisherige Veröffentlichungen des Indikators
Construction of the RWI-Consumption Indicator
Accounting for about 60% of GDP, private consumption is the most important component of aggregate demand in Germany. However, forecasting consumption has proved to be particularly difficult as the available leading indicators – such as the GfK consumer sentiment indicator or the index of retail trade turnover – are only weakly correlated with actual consumption.
Nowadays, the Internet has emerged as a new tool to obtain information on products and prices preceding consumption activities. According to a survey by the German Federal Statistical Office, 86 % of the Internet users in 2009 employed it as a source of information on products and services and 75% used it to purchase products and services. Search engines play a substantial role in gathering information on the internet as well as online-shopping. This applies in particular to Google’s search engine, which has a market share of around 90% in Germany.
The RWI-Consumption Indicator intends to utilize information on the web-search intensity provided by Google on a weekly basis. From the Insights for Search application aggregated indices of search queries can be obtained which are classified into more than 1000 categories and sub-categories on several levels using an automated classification engine. The Google time series are not subject to revisions but the data are based on random samples drawn each day by Google. This entails some variation between samples drawn on different days. The RWI-Consumption Indicator is therefore not constructed from the latest available time-series but from averages of the time-series that are drawn by the RWI every day. This procedure ensures both stability and reproducibility of the indicator. For the construction of the indicator, 45 consumption-relevant categories were selected that constitute best matches for the components of private consumption in the national accounts of the Federal Statistical Office. To avoid multiple usage of the same information, only series from either main categories or sub categories were used.
The Google time-series are available on a weekly basis back to 2004 and are not seasonally adjusted. They are first aggregated to monthly data using the simple averages and then seasonally adjusted by the CENSUS X12 ARIMA procedure. Next, a diffusion-index is computed from the monthly changes in the 45 time-series. This index displays the share of categories that increased in value relative to the previous month. Finally, the diffusion index which can take values between 0 and 100 is aggregated to a quarterly index.
Components of private consumption in national accounts and matching Google categories
SEA-No. and components according to national accounts / Google categories
01: Food and non-alcoholic beverages
Food and Grocery Retailers, Nonalcoholic Beverages
02: Alcoholic beverages and tobacco products
Alcoholic Beverages, Tobacco Products
03: Apparel and footwear
Apparel, Fashion Designers and Collections
04: Housing, water, electricity, gas and other fuels
Bed and Bath, Electricity, Home Insurance, Home Financing, Oil and Gas, Waste Management
05: Fixtures, household appliances
Home Appliances, Home Furnishings, Home Improvement, Homemaking and Interior Decor, Yard and Patio
06: Health care
Drugs and Medications, Health Insurance, Medical Facilities and Services
07: Traffic and Transportation
Freight and Trucking, Vehicle Brands, Auto Financing, Vehicle Parts and Accessories, Vehicle Shopping
Mobile and Wireless, Service Providers
09: Leisure, entertainment, culture
Animal Products and Services, Art and Entertainment, Book Retailers, Computer- and Video Games, Consumer Electronics, Ticket Sales, Newspapers, Photographic and Digital Arts, Toys, Travel Agencies and Services
11: Hotel and restaurant services
Hotels and Accommodations, Restaurants
12: Other products and services
Face and Body Care, Finance, Hair Care, Insurance, Social Services
Our investigations revealed that the RWI-Consumption Indicator leads consumption by 1 quarter. The index is thus useful not only for predictions of consumption in the current quarter but also providing information for the next quarter. The figure shows the RWI-Consumption Indicator and the growth rates of consumption between 2004:Q1 and 2012: Q1. To better visualize the co-movement between the two series, the RWI-Consumption Indicator values are displayed with a lag one quarter. Overall, the index predicted the dynamics of consumption fairly well. The distinct drop of consumption in 2007:Q1 is not reflected by a particularly low index value. However, the displayed tendency is still correct. The large drop in consumption of 2011:Q2 was also not indicated. However, the gap may shrink once the data have been finally revised. The experience shows that the index traces the finally revised growth rates of consumption much better than the first releases.
RWI (2010), Analyse und Prognose des Spar- und Konsumverhaltens privater Haushalte. RWI Projektberichte. download
Schmidt, T. und S. Vosen (2012), A Monthly Consumption Indicator for Germany Based on Internet Search Query Data. Applied Economics Letters 19 (7): 683-687.
Schmidt, T. und S. Vosen (2011), Forecasting Private Consumption: Survey-based Indicators vs. Google Trends. Journal of Forecasting 30 (6): 565-578.
Schmidt, T. und S. Vosen (2010), A Monthly Consumption Indicator for Germany Based on Internet Search Query Data. Ruhr Economic Papers #208. RWI. download
Schmidt, T. und S. Vosen (2009), Forecasting Private Consumption – Survey-based Indicators vs. Google Trends. Ruhr Economic Papers #155. RWI. download