Big Data: Google searches predict unemployment in EU-28 – ETLAnow


Commissioned by: ETLA, the Research Institute of the Finnish Economy 
Project duration: July, 2015 – January, 2016 
Project manager: Iva Tomić, PhD
Collaborator: Ivan Žilić



Brief outline: 
Project goals: 
ETLAnow is an economic forecast model. It provides daily updates on the current and future unemployment rate for every EU country, and publishes the forecasts automatically online. The model is based on the idea that volumes of Google searches on unemployment related matters, such as unemployment benefits, might be associated with the actual unemployment level. More generally, the main aim of the project was to use massive new real-time data sets, that is, big data, to improve economic forecasts.

Expected results/outputs: 
ETLAnow utilizes Google search data to predict the unemployment rate in every EU country. It predicts automatically the unemployment rate for three months ahead using data from Google Trends database and Eurostat, publishing the results on a daily basis. 

Activities: 
Activities of the researchers within the Institute of Economics, Zagreb primarily involved providing a list of Google search terms that an unemployed person, or a person expecting unemployment, would search for in Croatia. Additional activities included writing a paper/report on the forecasting performance of the system in Croatia with support from ETLA. 

Methodology: 
The model utilized real-time data on the volumes of unemployment-related Google searches as well as the latest official figures on the unemployment rates. Simple seasonal first-order autoregressive model, which included relevant Google variables (Google Index) was used for forecasting. In addition, cross-correlation analysis and Granger-causality tests were performed.
 

Vrh