摘要
在国家培育一流投资机构的政策背景下,科学评估基金管理能力已成为学术界、实务界及监管机构共同关注的核心议题。但现行依赖结构化数据的评价体系存在信噪比偏低、数据易受操纵、策略性隐藏等问题。为此,本文将传统基于“持仓行为—业绩表现”的基金评价逻辑前移至“市场认知”层面,并提出基于非结构化数据的度量方法,发展了基金能力评价框架。基于2008~2021年46050篇基金管理报告的文本数据,本文结合机器学习模型Doc2Vec,从管理报告意见的文本活跃度视角切入,衡量了基金的市场认知能力。研究发现:认知能力显著预测了基金的未来表现,并揭示了基金在择时能力上的差异,为现有主要侧重于捕捉择股能力的基金评价体系提供了重要补充。本文进一步从市场变化的异质性、认知到行为的转化路径以及投资者有限关注的外部影响这3个方面,探讨了该指标的预测机制,并验证了“认知—行为—表现”的逻辑通路。本研究丰富了现有的基金能力评价体系,有助于引导社会资金流向高能力基金,从而在宏观层面提升资金配置的市场效率,在微观层面改善中小投资者与社会保障性资本的长期福祉。
Against the backdrop of national policies aimed at cultivating world-class investment institutions,the evaluation of fund management capabilities has become a central concern shared by academia,industry,and regulatory authorities.However,the current evaluation system,which relies heavily on structured data,suffers from low signal-to-noise ratios,susceptibility to manipulation,and the potential for strategic concealment.To address these challenges,this paper shifts the traditional evaluation logic from"behavior-performance"to the dimension of"cognition",and proposes a novel assessment method based on unstructured data,thereby advancing the existing framework for evaluating fund capabilities.Drawing on a dataset of 46,050 fund management reports from 2008 to 2021,this study leverages the Doc2Vec machine learning model to measure market cognition capability from the perspective of textual activeness in fund manager commentaries.Moreover,it highlights dfferences in funds timing abilities,providing an important complement to the existing fund evaluation systems that primarily focus on capturing stock-picking skills.The paper further investigates the underlying predictive mechanisms of this indicator from three perspectives:the heterogeneity of market changes,the pathway from cognition to action,and the external impact of investors limited attention.These analyses provide clear empirical support for a"cognition-behavior-performance"causal chain.By enriching the current fund capability assessment system,this study offers valuable guidance for directing capital toward high-capability fund managers,thereby enhancing capital allocation efficiency at the macro level and improving the long-term welfare of retail investors and social security capital at the micro level.
作者
张维
陈卓
林兟
Zhang Wei;Chen Zhuo;Lin Shen(Laboratory of Computation and Analytics of Complex Management Systems(CACMS),Tianjin University;College of Management and Economics,Tianjin University)
出处
《管理世界》
北大核心
2025年第7期91-107,191,共18页
Journal of Management World
基金
国家自然科学基金(基金号:72371184、72141304、72342022)的资助。
作者简介
通讯作者:林兟。
