报告题目：Corporate News Prediction through News Insights
报告人：唐文金 博士（南方科技大学）
时间：2025年12月09日 10:30-11:30
地点：管理学院A北204
报告内容摘要：
This study investigates whether and how media news contains predictive information for first-time corporate bond defaults—a challenging and low-frequency event. We develop an interpretable and expert knowledge–driven text analysis framework that embeds financial expertise into textual modeling. Using a comprehensive dataset covering nearly all corporate bonds and media news in China from May 2015 to December 2022, we show that media news provides incremental and economically significant information beyond financial ratios and macroeconomic indicators, generating robust early-warning signals with horizons of at least six months. The empirical evidence demonstrates that, within this specialized financial context, the depth of embedded domain expertise outweighing both the breadth of general knowledge and complexity of the model. Our domain-informed dictionary consistently and robustly exceeds the performance of advanced alternatives, including Large Language Models, BERTopic models, and general-purpose textual lexicons. We further show that state-controlled and market-oriented media play complementary informational roles, differing in thematic focus across firm types yet both exhibiting selective disclosure toward central and state-owned enterprises, thereby revealing how institutional media structures shape information flows in credit markets.
【唐文金个人简介】
唐文金，北京航空航天大学博士，南方科技大学博士后，主要研究方向：行为金融、实证资产定价、金融科技、监管科技等。主持国家自然科学基金青年项目，参与国家自然科学基金重点项目、国家自然科学基金面上项目等多项课题研究，在SSCI和CSSCI国内外核心期刊上发表过多篇学术论文，多次在国内外学术会议上报告研究成果，并多次获得优秀论文奖。
欢迎有兴趣的教师，全体博士生、硕士生参加。
管理学院人才办公室
2025-12-05
