统计研究 ›› 2021, Vol. 38 ›› Issue (8): 30-44.doi: 10.19343/j.cnki.11-1302/c.2021.08.003

• • 上一篇    下一篇

基金业绩分析——基于有向学习网络的研究

罗荣华 赵森杨 方红艳   

  • 出版日期:2021-08-25 发布日期:2021-08-25

Mutual Fund Performance Analysis:Based on Directional Learning Network
#br#

Luo Ronghua Zhao Senyang Fang Hongyan   

  • Online:2021-08-25 Published:2021-08-25

摘要: 本文使用LASSO算法构建了基于基金持股数据的基金间动态学习网络,将基金研究中传统的无向网络扩展为有向网络,并检验了正向学习与反向学习两种不同的学习模式(信息利用方式) 对基金业绩的影响,进而探讨了其背后的经济含义。实证结果表明:当基金作为被学习者(信息被观测方)时,被正向学习会显著提高其业绩,被反向学习会显著降低其业绩;当基金作为主动学习者(信息观测方)时,无论是正向学习还是反向学习均不会对其业绩造成显著影响;对基金学习动机的分析表明,基金参与学习是为了提升相对自己上期的业绩、防止业绩倒退,且反向学习相对更加有效。本文分析了信 息传递方向、信息利用方式对基金业绩的影响,为如何将统计学习方法应用于金融问题的分析提供了一个新的视角。

关键词: LASSO, 基金业绩, 学习网络

Abstract: Based on mutual fund portfolio holdings data, we build dynamic learning networks of mutual funds using LASSO, which extends traditional unidirectional fund networks to directional networks. We then analyze the impact of different learning modes, specifically, concordant learning and opposite learning, on mutual fund performance, and explore the economic implications of these learning modes. The empirical evidence shows that:Firstly, funds that are learned by others in the concordant mode can result in a performance improvement, while performance deteriorates if they are learned by others in the opposite mode. Secondly, funds, as learners, acting either in the concordant or the opposite direction, do not experience significant changes in performance. Finally, the analysis of the learning motivation shows that funds participate in learning to improve the performance from the previous cycle and avoid performance deterioration that may occur otherwise, and opposite learning is more effective. Our paper contributes to the literature by exploring, for the first time, the impact of the direction of information transfer and the way information is used on fund performance, and presents new insight into how statistical learning methods can be applied in financial research.

Key words: LASSO, Mutual Fund Performance, Learning Network