This paper examines how credit sentiments in earnings conference calls affect investors in the the stock market and debt market. Credit sentiments refer to the tone of the management and analysts when they discuss the credit situation of the firm during the conference calls. We use an advanced natural language processing (NLP) technique—FinBERT (Huang, et al. 2021) for our sentiment analysis. Credit discussions account for less than 20% of the conference calls but they contain important and timely information about the firms’ credit risks, which precedes the rating changes made by credit rating agencies (CRAs). Bond investors, especially retail investors, traditionally rely on the ratings from credit rating agencies to evaluate the credit risks of firms. However, the credit ratings agencies have been usually criticized for failing to update their credit ratings on a timely basis. Our findings show that more positive credit sentiments in conference calls are associated with lower borrowing costs, more positive credit rating changes, lower CDS spread as well as lower likelihood of bankruptcies and defaults in the future. The stock market reacts positively to credit sentiments around earnings announcement while the bond market under-reacts to this information and generates a post earnings announcement drift. Evidence suggests that the under reaction from the bond market is mainly due to retailed investors. Bond trading strategy based on credit sentiments in conference calls generates significant returns for bond investors.
Bio: I am a fourth year accounting PhD student at Haas School of Business. My research interest lies in textual analysis of firms' voluntary disclosures to study how they affect the investors and analysts in the capital market.