基于贝叶斯推断的证券投资基金绩效分析  

Evaluating Fund Performance Based on Bayesian Inference

作　　者：朱洪亮[1] 陈莹[1] 石韵青[1] 刘匡民[1] 

机构地区：[1]南京大学工程管理学院

出　　处：《管理学报》2012年第7期1013-1019,共7页Chinese Journal of Management

基　　金：国家自然科学基金资助重点项目(70932003);国家自然科学基金资助项目(70701016;71171109;71173098)

摘　　要：从投资者先验信念的角度,运用贝叶斯推断对我国封闭式基金绩效进行实证研究。通过直观的问题引出投资者对基金管理者技巧的先验信念集合,并与具有4个风险指标的线性模型相结合,有效联系了投资决策过程中的重要因素——个人直观理念与实际市场数据,给出投资者在不同先验信念下的后验绩效。研究表明,随着投资者对管理者先验信念的增加,基金后验绩效增加,投资者更倾向于投资;只有当投资者对管理者有极强的先验信念时,才会投资于表现一般的基金;同理,不投资于绩优基金也需极强的先验信念。同时发现,基金费用对基金后验绩效的影响主要反映在基准线的平移。进一步,通过分时间段的投资组合权重分析,给出投资者在不同市场环境下的最优投资组合策略选择方法。The performance of the Chinese close-end funds is empirically studied by Bayesian infer-ence from the investors＇ prior believes. Portfolios problem of an investor is analyzed ana me posterior performance evaluation with a four-factor model and a series of prior believes is brought in, which enables the connection between the prior believes which are given by intuitive questions of the investors and the market data. The results show that with the increasing of the prior, the posterior performance would increase and the investors are more likely to invest, and that the investor would not invest a general fund unless he has strong prior on the manager and strong prior is also needed when he is convinced not to invest the good-performance fund, at the same time by analyzing the weight of each securities in different period, the optimal portfolios of different period could be given.

关 键 词：证券投资基金 贝叶斯推断 绩效分析 投资组合 

分 类 号：C93[经济管理—管理学] F830.91