Prompt Design for Monetary Policy: An LLM-Based Sentiment Analysis for Macroeconomic Decision-Making

Central banks use monetary policy for maintaining economic stability. To enhance economic growth, central banks need to understand market sentiment, as it can influence the economic performance. Recent developments in Large Language Models (LLM) demonstrated the potential for more informed and comprehensive sentiment analysis. This paper explores the potential application of LLM in enhancing monetary policy decision-making through sentiment analysis of textual data. To evaluate the accuracy of LLM in identifying sentiment from financial news and monetary policy, this research conducts data collection, prompt design, and sentiment analysis. The results show high accuracy for different types of LLM prompts. A strong correlation was found between financial news sentiment and inflation rates which enables the construction of inflation rates predictive model based on sentiment data. The findings suggest that LLM can provide central banks with a deeper understanding of market sentiment for more proactive and informed economic decisions.