Over 1 billion euros spent in foreign EU webshops

In 2016, Dutch consumers spent over 1 billion euros on products sold by foreign EU webshops. This represents an increase of 25 percent relative to 2015, as reported by Statistics Netherlands (CBS) on the basis of current big data research in cooperation with the University of Amsterdam and Leiden University.

[video: https://www.cbs.nlen-gb/video/6dd8bf5b70394b3ba13b1e187c662205] 
(English subtitles available)

More than half of all online purchases were made at webshops located in Germany. Clothing and shoes were the main items purchased by consumers.
These purchases concern online purchases of goods by Dutch consumers from companies located within the European Union, but outside the Netherlands. Figures have been calculated using a method which is still under development and are therefore provisional. The margin of error in relation to total turnover amounts to approximately 5 percent.

Substantial growth in total spending

Dutch consumer spending in foreign European webshops grew rapidly between 2014 and the end of 2016. Expenditure rose from 660 million euros in 2014 to 860 million euros in 2015 and continued to grow to more than 1 billion euros in 2016.
In 2016, expenditure was up by 62 percent relative to 2014. In comparison: the volume of purchases from Dutch webshops increased by 45 percent between 2014 and 2016.

Online expenditure at foreign webshops vs. total retail expenditure

The Dutch retail sector recorded a total turnover of 107.1 billion euros in 2015. This means that the value of Dutch consumer spending in foreign European webshops accounted for 0.7 percent of total retail turnover in that year. Foreign online purchases are not included in Dutch retail figures but the above percentage does provide insight into total consumer spending and is an indicator of competition for the Dutch retail sector from foreign webshops.

Most purchases from Germany

Last year, over 80 percent of online purchases by Dutch consumers in other EU countries were from companies located in Germany, Great Britain, Belgium or Italy. German webshops topped the list with a share of more than 50 percent in total related purchasing value. Great Britain ranked second at around 12 percent, followed by Belgium and Italy at approximately 8 percent each. The other 20 percent was spent in webshops from across the entire Community. The corporate location is central to establishing these figures rather than the domain name or website language, which in many cases is Dutch.

Mainly clothing and shoes

Dutch consumer purchases from foreign EU webshops are mostly at shops selling clothing and shoes. The latter accounted for nearly 60 percent of the total online purchasing value related to foreign EU webshops in 2016. In addition, consumers spent their money mainly in shops selling animal nutrition products, home furnishing articles and food supplements; purchases were also made in other sectors. Fifteen percent of purchases could not be clearly traced back to any product category.

Purchases from EU webshops by type
 20162014
Clothing and shoes5852
Animal nutrition78
Home furnishing33
Food supplements21
Other1416
Unknown1520

Method developed in collaboration with the University of Amsterdam and Leiden University

To measure expenditure on non-Dutch webshops located in the EU, Statistics Netherlands (CBS) uses the Dutch VAT returns filed by foreign EU retailers. The challenge is to identify the webshops from among this data.

The method for identification of webshops from among the VAT returns has been developed in collaboration with the University of Amsterdam and Leiden University. The University of Amsterdam has access to Dutch Chamber of Commerce registrations of around 90 million European companies. These have been classified according to the Standard Industrial Classifications (SBI) 2008. This classification is used by CBS and all other national statistical institutes of the European Union for the publication of official economic statistics. The registrations include approximately 3 million foreign EU retail businesses, among which are the foreign EU webshops. Generally speaking, registrations are well-covered across the European Union.

The challenge is linking the Chamber of Commerce registrations from elsewhere in Europe to the Dutch VAT returns. Linking needs to be done by company name; however, a company may be represented under different names in these two sources. For example, there may be differences in punctuation marks (dots, commas, dashes etc. ) as well as abbreviations (e.g. LTD versus LIMITED). Moreover, comparing the names takes up a great deal of calculation time in such large datasets. In this survey, CBS together with the University of Amsterdam and Leiden University implemented existing advanced research techniques in order to achieve both faster and more accurate linking. This resulted in an overview of all foreign EU companies filing VAT returns in the Netherlands which also have been registered as foreign EU retailers with the Dutch Chamber of Commerce.

Checks and measurement errors

To check the results from this linking process, verification takes place in two steps. First, internet data are collected by webscraping to check for websites of these shops through which products can be ordered online. Webpages are found on the basis of the company name, after which the pages are checked for the display of a shopping cart.

The next step is manual checking for webshop features in the websites of the largest foreign companies in terms of turnover size in the Netherlands. The product group related to these webshops is established with certainty. Through these manual checks, a rough estimate can be made of measurement errors in the appropriate algorithm. The resulting margin of error in item turnover is set at approximately 5 percent. Figures on annual growth in expenditure also have an error margin of around 5 percent. The latter growth figure had an error margin of more or less 8 percent for expenditure between 2014 and 2016 at 62 percent. Therefore, this growth figure should be between 54 and 70 percent. With the help of manual check results, the next version of the algorithms can be ‘trained’ using machine learning in order to further reduce measurement errors.

Follow-up survey

In the coming period, CBS intends to study the possibility of quarterly publications on Dutch consumers’ online purchases from foreign EU webshops.
In addition, CBS will conduct part two of the survey in the fall of this year; this will focus on Dutch consumers’ online purchases from webshops outside the EU. Non-EU companies do pay taxes in the Netherlands, but the VAT declaration is often transferred to Dutch transporters. CBS has access to data available at Dutch customs authorities from which data on such declarations may be derived.

Center for Big Data Statistics

This survey was conducted at the Center for Big Data Statistics (CBDS), established by CBS in September 2016. At the CBDS, around 40 collaborative partners from the public and private sector, education and science work on implementing big data for the production of official statistics. The CBDS focuses on improving the timeliness of statistics, starting the production of new statistics and adding more depth to existing statistics, e.g. by breaking these down according to region or sector of industry. This will lead to better policy support, whilst automated integration of data sources and flows can offer considerable efficiency improvement.
In consultation with its partners, the CBDS is working on a wide variety of projects. This survey is one such project. Results derived from CBDS projects are published on the CBS innovation pages.

You are using an unsupported browser. Please upgrade your browser. You provided an inconsistent user agent while solving the challenge. You may have browser extensions or settings enabled to spoof the user agent and should disable them to proceed. Some portion of Turnstile was accidentally cached. Please clear your cache. The time on the clock is incorrect. Please set your clock to the correct time. An unspecified error occurred.