Wykorzystanie Google Trends do modelowania stopy bezrobocia rejestrowanego w Polsce
Use of Google Trends in modelling registered unemployment rate in Poland

Summary/Abstract: The paper deals with the problem of monitoring the unemployment rate in Poland. The main aim of the article is to check whether the addition of selected Google Trends indices improves the accuracy of forecasts generated by the autoregressive model of registered unemployment rate. The research method is based on nowcasting techniques which are used to assess the current state of an economy. Data for the years 2004–2019 were retrieved from publication by Statistics Poland (GUS) and Google Trends, the latter of which allows tracking the popularity of terms searched by Internet users. The study compares the goodness of fit and forecast errors of the basic model with these of models extended with exogenous variables. Both the potential and the limitations of the utilisation of a new source of data in macroeconomic analyses concerning Poland are presented in the paper. The analysis yields a conclusion that Google indexes (commonly used in the literature written in English) do not improve the accuracy of predictions of the autoregressive model. Satisfactory results are only obtained for indices related to the international mobility of the workforce.

• Details
• Contents