【题目】： Textual Sentiment and the Cross-Section of Stock Returns
【主讲人】：陈赟 北京大学国家发展研究院 博士研究生
【时间】：2019年11月29日（周五）10:00-11:30
【地点】：北大经济学院101室
【摘要】：This paper uses internet message data to measure firm-level textual sentiment and examine its predictability on stock return. We find that the stocks with high abnormal daily close-to-close textual sentiment generate statistically significant, positive, and economically larger abnormal returns than the low-sentiment stocks, even after controlling for firm characteristics and common pricing factors. These abnormal returns are stronger among stocks with a larger number of messages, smaller size, lower EP ratio, higher illiquidity, higher idiosyncratic volatility and extreme returns. We next explore the source of this predictability and find that the effect of abnormal textual sentiment on returns will not completely reverse within a year, that the abnormal textual sentiment positively predicts firm's earnings surprises. We also find that the predictability of abnormal message sentiment is stronger for firms with media news, and it survives after controlling the media tone. These findings are consistent with the hypothesis that internet short messages contain the firms' fundamental information.
【主讲人简介】：
陈赟，北京大学国家发展研究院金融学博士生
2018至2019年访问美国杜克大学经济系
他的研究领域包括文本大数据分析、实证金融、机器学习、金融科技等。
