Forecasting GDP in Europe with Textual Data


We evaluate the informational content of news-based sentiment indicators for forecasting the Gross Domestic Product (GDP) of the five major European economies. The sentiment indicators that we construct are aspect-based, in the sense that we consider only the text that is related to a specific economic aspect of interest. In addition, the sentiment is fine-grained as each word is assigned a score in the interval [-1, 1]. Our data set includes over 27 million articles for 26 major newspapers in 5 different languages. The evidence indicates that these sentiment indicators are significant predictors to forecast GDP and their predictive content is robust to controlling for macroeconomic and survey confidence indicators available to forecasters in real-time. We also discuss the application of the sentiment indicators during the COVID-19 pandemic and demonstrate their relevance in nowcasting GDP.

Journal of Applied Econometrics