Abstract
This paper shows that newspaper articles contain signals that can materially improve real-time nowcasts of real GDP growth for the Euro area. Using articles from 15 popular European newspapers, which are machine translated into English, we create sentiment metrics that update daily and assess their value for nowcasting, comparing with competitive and rigorous benchmarks. We find that newspaper text is especially helpful early in the quarter before other indicators are available. We also find that general-purpose sentiment measures perform better than more economics-focused ones in response to unanticipated events and nonlinear supervised models can help capture extreme movements in growth but require sufficient training data to be effective.
| Original language | English |
|---|---|
| Pages (from-to) | 887-905 |
| Number of pages | 19 |
| Journal | Journal of Applied Econometrics |
| Volume | 39 |
| Issue number | 5 |
| Early online date | 1 May 2024 |
| DOIs | |
| Publication status | Published - Aug 2024 |
JEL classifications
- c43 - "Index Numbers and Aggregation; Leading indicators"
- c45 - Neural Networks and Related Topics
- c82 - "Methodology for Collecting, Estimating, and Organizing Macroeconomic Data; Data Access"
- e37 - Prices, Business Fluctuations, and Cycles: Forecasting and Simulation: Models and Applications
Keywords
- business cycles
- COVID-19
- forecasting
- machine learning
- text analysis
- FACTOR MODELS
- BIG DATA