The Power of Text-based Indicators in Forecasting the Italian Economic Activity

Abstract

Can we use newspaper articles to forecast economic activity? Our answer is yes and, to this end, we propose a brand new economic dictionary in Italian with valence shifters, and we apply it to a corpus of about two million articles from four popular newspapers. We produce a set of high-frequency text-based sentiment and policy uncertainty indicators (TESI and TEPU respectively), which are constantly updated, not revised and computed both for the whole economy and for specific sectors or economic topics. To test the predictive power of our text-based indicators, we propose two forecasting exercises. First, by using Bayesian Model Averaging (BMA) techniques, we show that our monthly text-based indicators greatly reduce the uncertainty surrounding the short-term forecasts of the main macroeconomic aggregates, especially during recessions. Secondly, we employ these indices in a weekly GDP growth tracker, achieving sizeable gains in forecasting accuracy in both normal and turbulent times.