主管：中国科学院
主办：中国优选法统筹法与经济数学研究会
　　　中国科学院科技战略咨询研究院
中国管理科学

中国管理科学 ›› 2019, Vol. 27 ›› Issue (12): 1-10.doi: 10.16381/j.cnki.issn1003-207x.2019.12.001

• 论文 •    下一篇

基于期望效用-熵模型的基金评级方法及其在中国基金评级中的应用

杨继平1, 石晨晓1, Daniel CHIEW2, Judy QIU2, Sirimon TREEPONGKARUNA2   

  1. 1. 北京航空航天大学经济管理学院, 北京 100191;
    2. 西澳大利亚大学商学院, 珀斯 6009
  • 收稿日期:2018-08-01 修回日期:2018-09-30 出版日期:2019-12-20 发布日期:2019-12-30
  • 通讯作者: 杨继平(1963-),男(汉族),内蒙古呼和浩特人,北京航空航天大学经济管理学院教授,博士,研究方向:风险决策、金融计量建模等,E-mail:yangjp@buaa.edu.cn. E-mail:yangjp@buaa.edu.cn
  • 基金资助:
    国家自然科学基金资助项目（71271011，71571009）

A Fund Ratings Approach Based on EU-E Decision Model and Its Application in China's Fund Ratings

YANG Ji-ping1, SHI Chen-xiao1, Daniel CHIEW2, Judy QIU2, Sirimon TREEPONGKARUNA2   

  1. 1. School of Economics and Management, Beihang University, Beijing 100191, China;
    2. Business School, University of Western Australia, Perth 6009, Australia
  • Received:2018-08-01 Revised:2018-09-30 Online:2019-12-20 Published:2019-12-30

PDF (PC)

189

摘要： 基金评级对于投资者来说具有重要的参考价值，研究合适的基金评级方法非常必要。本文针对晨星评级对风险调整和预测能力不足的特征，研究应用期望效用-熵（EU-E）模型基金评级方法对我国开放式基金进行评级的预测能力；并以Sharpe指数、Jensen、Fama-French三因素和Carhart四因素α作为业绩指标，利用固定效应面板数据回归模型对期望效用-熵模型和随机效应面板数据回归模型对晨星基金评级的预测能力进行比较分析。采用样本期由2011年2月到2016年6月的261只基金为研究样本进行评级；研究结果表明，基于期望效用-熵平衡系数λ=0.25和0.75时，EU-E模型基金评级方法评级具有良好的预测能力，而晨星评级预测能力较弱。特别地，λ=0.25和0.75时，EU-E模型评级的五星级基金业绩优于晨星评级对应的基金业绩，而且相比于晨星评级可以更好地区分不同星级基金的业绩。另外，研究结论对于短期、中期和长期的样本都是稳健的。

关键词: 基金评级, 开放式基金, EU-E模型, 晨星评级, 预测能力

Abstract: Due to demand and growth in the fund industry, simple fund rating tools as a measure of fund performance are becoming more and more important to assist investors in making capital allocation decisions. Thus, it is essential to develop an appropriate approach to rank funds. Previously, Yang and Qiu developed the expected utility-entropy (EU-E) decision model, which brings together the notion of expected utility and entropy together and can effectively consider the decision maker's subjective preferences and objective uncertainty at each state of nature. In the paper, we investigated an alternative fund rating approach based on EU-E model to mitigate drawbacks of the Morningstar ratings, and the ability of fund ratings based on EU-E model and Morningstar to predict fund performance. The Sharpe index, Jensen α, Fama-French three factors and Carhart four factors α have been used as the performance indices using the panel data regression with fixed effect model and random effect model respectively according to the results of F test and Hausman test. The sample period is selected over February 2011 to June 2016, and 261 of funds in the sample are included. The results show that the fund ratings based on EU-E model when λ takes value of 0.25 and 0.75 have excellent forecasting ability while Morningstar ratings have weak forecasting ability. In particular, the 5-star fund performance based on EU-E model when λ takes 0.25 and 0.75 is better than that of the Morningstar ratings; and the EU-E model can better distinguish the performance of different star funds. Furthermore, the conclusions are also robust to one-, three-, and five-year sub-samples. The established approach can provide an additional way to evaluate fund performance and further assist retail investors to better investment decision making.

Key words: fund rating, open-end funds, EU-E model, Morningstar ratings, predictive ability

中图分类号: