Improving unemployment rate forecasts at regional level in Romania using Google Trends
2020; Elsevier BV; Volume: 155; Linguagem: Inglês
10.1016/j.techfore.2020.120026
ISSN1873-5509
Autores Tópico(s)Nutritional Studies and Diet
ResumoThe availability of online data in real time has the potential to explain and predict better the economic indicators for which official data are provided with delay. Considering the advantages of Google Trends in providing Internet data, the aim of this paper is to explain and predict the regional unemployment rate in Romania at county level. More panel data models were constructed for the period 2004–2018, some of these including only variables from official statistics and other models combining Google Trends data for keywords like locuri de muncă, joburi and angajări (jobs and hiring) with macroeconomic indicators from National Institute of Statistics. Moreover, Granger causality was checked between unemployment and the other indicators. Compared to previous studies in this field, this paper brings the regional approach in explaining unemployment rate and the use of panel data models. Predictions for unemployment rate at county level for 2018 based on a panel data model with Internet data and official data outperformed those forecasts based only on official indicators. The results suggest that indicators collected through Google Trends might improve the unemployment rate in Romania and should be considered in providing better forecasts to support the government decisions. However, our results could not be generalized since forecast accuracy is conditioned by the stability of the constructed Internet variables.
Referência(s)