媒体情绪与企业风险承担——基于机器学习和文本分析的证据

Media sentiment and corporate risk-taking——Evidence based on machine learning and text analysis

媒体舆论作为企业运营过程中与外界信息交互的重要渠道,能够通过信息加工和反馈潜移默化地影响企业决策.基于优化机器学习方法测量得到的媒体情绪得分,本文发现媒体情绪会抑制企业风险承担,这一结论经过充分稳健性检验后仍然成立.机制分析表明,媒体情绪主要通过影响信息不对称程度、感染投资者情绪和加剧分析师羊群行为实现对企业风险承担的抑制效应.异质性分析表明相比于负面媒体情绪,正面媒体情绪在这一过程中起主导作用.除此之外,这一抑制效应在行业竞争程度更激烈、外部监督程度更严格时显著性降低,且在融资约束更高、金融化程度更高、高管在职消费现象更严重时更加明显.本文的研究不仅有助于厘清媒体情绪对企业风险承担的影响方式和途径,对于如何利用媒体力量助力企业风险决策和价值提升,促进实体经济高质量发展也有一定的政策启示.

As an important channel of information interaction with the outside world in the process of enterprise operation,media opinion can influence enterprise decision-making imperceptibly through information processing and feedback.Based on the media sentiment scores measured by optimized machine learning methods,this paper find that media sentiment can inhibit corporate risk-taking.This conclusion still holds true after sufficient robustness testing.Mechanism analysis shows that media sentiment can inhibit corporate risk taking mainly by affecting the degree of information asymmetry,infecting investor sentiment,as well as exacerbating analysts’herding behavior.Heterogeneity analysis showed that positive media sentiment played a dominant role in this process compared to negative media sentiment.In addition,this inhibitory effect is more significant when the degree of industry competition is more intense and the degree of external supervision is stricter.Simultaneously,it is less obvious when the financing constraints are higher,the degree of financialization is higher and the phenomenon of perquisite consumption by executives is more serious.The research in this paper not only helps to clarify the ways and means by which media sentiment affects corporate risk-taking,but also has policy implications for how to use media power to help corporate risk decision-making and value enhancement,and to promote high-quality economic development.