Impact of Firm-specific Textual Sentiment on Credit spreads

While the impact of investor sentiment on the stock market has been extensively studied, there remains a notable gap in understanding similar dynamics in the bond market. To address this gap, this study examines the relationship between credit spreads and textual sentiment. Using natural language processing, we extract firm-specific sentiment expressed in news articles and blog posts and examine it them based on textual tone and emotion. Our findings are as follows. First, we find that sentiment is positively related to the subsequent decline in credit spreads. Sentiment derived from the news is more predictive of credit spreads than the sentiment derived from the blogs. Notably, the relationship between sentiment and credit spreads varies across market types. There is a negative correlation between sentiment and the following month's credit spread for KOSPI-listed firms, while a positive correlation is observed for KOSDAQ-listed firms. During crisis periods, such as the US-China trade war and COVID-19, the impact of sentiment on credit spreads increases. A long-short portfolio strategy based on sentiment generates significant profits, confirming the economic importance and the potential for investors to use sentiment analysis to generate returns.