Publié le 8 janvier 2025 Mis à jour le 15 janvier 2025
Lieu(x)
Pôle Tertiaire - Site La Rotonde - 26 avenue Léon Blum - 63000 Clermont-Ferrand
Salle 215

Séminaire recherche. Analyzing Central Bank Communications Using Machine Learning Tools

Romain Veyrune
Fonds monétaire international 

Abstract 

The presentation provides an analysis of the role of central bank communications and the development of an automated classification tool to assess their effectiveness. Central bank communications are crucial for shaping market expectations and influencing economic decisions. The study utilizes a dataset comprising over 24,000 documents from more than 100 central banks, focusing on various metrics such as topic, stance, audience, and sentiment. The analysis identifies challenges in translating complex textual data into measurable metrics, underscoring the necessity for a robust classification framework. The automated tool categorizes communications at a sentence level, enabling insights into how different communication styles impact economic conditions, particularly inflation volatility. Empirical results indicate that net policy sentiment correlates significantly with policy rates, showcasing the importance of communication strategies. The study also reveals a U-shaped relationship between net policy stance and inflation volatility, emphasizing the predictive power of communication metrics in economic forecasting. Ultimately, the findings suggest that effective central bank communication can enhance credibility and stabilize economic conditions, with further research needed to explore additional data sources and their implications for economic outcomes. This innovative approach marks a significant contribution to understanding central bank communications' impact on the broader economic landscape.