Treasury yield spread prediction with sentiments of Beige Book and macroeconomic data

The yield spread is a crucial factor that affects both the economy and financial markets, and it is greatly influenced by decisions made by the Federal Open Market Committee (FOMC). To make these decisions, the FOMC relies on the Beige Book, which provides a comprehensive overview of current economic conditions and trends. As a result, yield spread fluctuations respond to the FOMC’s decisions at the end of their meetings. In this paper, we propose a treasury yield spread prediction method that uses the XGBoost model. We combine sentiment data analyzed by the FinBERT model with macroeconomic data to produce more accurate predictions. Our results show that by using both macroeconomic and sentiment data, we can improve the accuracy of yield spread predictions from the release of the Beige Book until the end of the FOMC meeting. Furthermore, we find that focusing on the sentiment regarding the Fed’s Dual Mandate for inflation and employment in the Beige Book leads to more accurate predictions. Specifically, this means that sentiment analysis of economic conditions related to inflation and employment can provide valuable information for predicting yield spread fluctuations.