CMStatistics 2020: Start Registration
View Submission - CFE
Title: Nowcasting GDP growth using media news articles Authors:  Arno De Block - Vrije Universiteit Brussel (Belgium) [presenting]
Andres Algaba - Vrije Universiteit Brussel (Belgium)
Geert Langenus - National Bank of Belgium (Belgium)
Peter Reusens - National Bank of Belgium (Belgium)
Abstract: GDP growth is the key indicator for measuring the state of the economy. It is closely monitored by policymakers, firms, and investors to optimize their economic decision-making. But while decisions have to be made in real-time, GDP growth is only observed at a quarterly frequency and with a substantial publication lag. To monitor the state of the economy in real-time, high-frequency macroeconomic and financial variables are used to nowcast GDP growth. We propose to augment this real-time dataset with information embedded in media news articles. The regression-based nowcasting approaches allow us to construct media-based variables that are optimal for the application of nowcasting GDP growth. Moreover, by enforcing sparsity, we manage only to select the most informative macroeconomic, financial, and media-based variables leading to superior nowcasting performance.