Under its partnership with the Banque de France, the Federation of E-Commerce and Distance Selling (Fédération du e-commerce et de la vente à distance - FEVAD) has provided monthly consumer online retail sales data since 2012. Pending the release of new data, the Banque de France carries out estimations, a task complicated by the growth of online retail. The autoregressive model (SARIMA(12)) used up to now can now be complemented by other statistical models that draw on exogenous data with a longer historical time series. This paper details the system of choices that results in the final forecast: data conversion, variable selection methods and forecasting approaches. In particular, Google queries, as measured by Google Trends, help enhance the predictive accuracy of the final model, obtained by combining single models.

Article on one page

To cite this article

Robin, F. (2018). Use of Google Trends Data in Banque de France Monthly Retail Trade Survey. Economie et Statistique / Economics and Statistics, 505-506, 35–63.
https://doi.org/10.24187/ecostat.2018.505d.1965