Platform data in accommodation tourism

Over the past few years, several apps have been in use that connect people to exchange goods and services. More and more consumers are using an online platform to reserve a holiday home or have a home meal delivered by a bicycle courier company. As a result, the economic importance of the collaborative economy is rapidly increasing. This is why Statbel, the Belgian statistical office, in close cooperation with Eurostat and other national statistical institutions, is studying how the collaborative economy can be integrated into public statistics. However, in the analysis of the platform companies, national statistical institutions are faced with a considerable difficulty. The largest platform companies are multinational players, who often manage their activities in Belgium from a foreign seat. These enterprises are therefore rarely found in the regular business statistics or registers. In order to obtain the required data, national statistical institutions would therefore be obliged to contact all platform companies on a unilateral basis. This was a time-consuming and inefficient way of working, both for the platform companies and for the statistical institutions. That is why the European Commission decided to attract these talks and to request the dates for all EU Member States through a single agreement. These negotiations focused primarily on the tourism industry and resulted in agreements with the platform companies Airbnb, Booking.com, TripAdvisor and Expedia. In the meantime, these companies have provided the first data files to Eurostat. Eurostat then divides the microdata into 27 national pseudonymised and aggregated files, providing Statbel with information on all reservations and overnight stays reserved on these four online platforms on Belgian territory. With the agreements between the European Commission and the four platform companies, a first, important hurdle was taken. But the methodological work begins only now. On the basis of the first stocks, national statistical institutions and Eurostat still need to develop a harmonised approach to methodological challenges. In particular, due to the fact that the microdata of the platform companies do not contain identification data, double counting poses a significant problem. These double countings, which include an accommodation in at least two different stocks, are above all a challenge to determine the capacity. That is why this information is not included in experimental statistics. At present, the national statistical institutes, together with Eurostat, are studying which techniques can best be used to solve the methodological problems. The focus is on innovative methods such as webscraping. Webscraping scrapes relevant information from websites, which is seen in combination with artificial intelligence as the best solution. In concrete terms, we study the following two slopes: text recognition: individuals who offer the same room on multiple online platforms generally use the same text. By searching for keywords, such as the location of the accommodation, the size of the room, facilities available, etc., identical accommodations can be automatically detected; photo recognition: this technique automatically compares the photos placed with an advertisement, in order to identify any double counting. However, this technique requires a large computer memory and is more often used as an alternative solution. In the long run, the aim is to integrate the platform data into the recurrent statistics. The timing for this depends both on achieving a harmonised approach to methodological problems and on faster data delivery by platform companies.