THE SENTIMENT ANALYSIS OF MPC MINUTES OF THE MEETINGS AND RELATIONSHIP WITH BOND YIELD AND SET INDEX

Abstract

With the increasing application of textual analysis tool to quantify qualitative data, this study aims to contribute to the academic discussion on textual analysis of the meeting minutes of the Monetary Policy Committee of the Bank of Thailand by examining the effectiveness of sentiment analysis tool in determining the net tone of the document and how such indicator relates to the common financial market indicators such as bond yields and the SET Index. By applying Valence Aware Dictionary for sEntiment Reasoning, this study found that the net tone derived from the compound values of each MPC minutes of the meeting was consistent with key economic events for both expansion (positive sentiment) and slowdown (negative sentiment). To validate relationship with Bond Yields and SET Index,  Vector Auto Regressive (VAR) has been applied to validate the relationship between the net tone and 1 year (1Y), 2 year (2Y), 5 years (5Y) and 10 years (10Y) government bond yields and the SET Index. Based on VAR, the net tones were significantly correlated with 10 year bond yields. Relationship with shorter maturities and the SET Index was not supported.