JEL Codes: G12, G14, G15
Keywords: news sentiment, bond risk premia, interest rates, factor analysis, machine learning, word-to-vector
학회소식
[2022년 5차 CAFM2022]News Sentiment and Bond Risk Premia
작성자 : 관리자
조회수 : 337
게시일 :
2022-12-20
This paper examines whether the local news sentiment for interest rate changes predicts bond risk premia in emerging markets where sovereign bonds, as benchmark securities, play a crucial role in the development of financial markets. Using machine learning techniques, we quantify the sentiment of 13,255 interest-rate-related articles published in three major Korean economic news media. In Korean treasury bond markets, the news sentiment indicating interest rates rises predicts lower bond risk premia over the subsequent 12 months. We construct a single return forecasting factor for bond risk premia based on the level, change, and square value of monthly average news sentiments. The news sentiment factor has predictive power after controlling for previously documented forecasting factors—forward rates, macroeconomic variables, and investor sentiments. The excess return predictability is stronger for longer tenor bonds.
첨부파일
