摘要
本文在Loughran and MacDonald(2011)词典的基础上通过人工筛选和word2vec算法扩充,构建了一个更新更全面的中文金融情感词典。我们使用该情感词典计算我国财经媒体文本情绪指标,发现媒体文本情绪可以更准确地衡量我国股市投资者情绪的变化,对我国股票回报有显著的样本内和样本外预测能力。媒体文本情绪对一些重要的宏观经济指标也有显著的预测能力,具有重要的学术和实践应用价值。
We construct a novel Chinese financial sentiment word dictionary based on the Loughran and MacDonald(2011)dictionary,word2 vec algorithm,and hand collection.The media textual sentiment calculated by this dictionary captures changes in investor sentiment well,and has significant predictive ability for Chinese stock market returns both in and out of sample,which is greater than the predictive power of commonly used macroeconomic indicators.In addition,textual sentiment is also a significant predictor for macroeconomic condition.
作者
姜富伟
孟令超
唐国豪
FUWEI JIANG;LINGCHAO MENG;GUOHAO TANG(Central University of Finance and Economics;Peking University;Hunan University)
出处
《经济学（季刊）》
CSSCI
北大核心
2021年第4期1323-1344,共22页
China Economic Quarterly
基金
国家自然科学基金(72072193、72003062、71872195)
中央财经大学创新群体项目资助。
关键词
媒体文本情绪
情感词典
收益预测
media textual sentiment
sentiment dictionary
return predictability
