基于弹性网分位数回归的开放型基金绩效研究  

Open-end Fund Investment Performance Based on Elastic Net Quantile Regression

作　　者：王文胜 宋家辉 WANG Wen-sheng;SONG Jia-hui(School of Economics,Hangzhou Dianzi University,hangzhou 310018,China)

机构地区：[1]杭州电子科技大学经济学院,浙江杭州310018

出　　处：《数理统计与管理》2020年第4期721-733,共13页Journal of Applied Statistics and Management

基　　金：国家自然科学基金项目(11671115)。

摘　　要：开放型基金是证券投资业务的重要形式之一,其风格识别和绩效评价对投资者来说都有很好的借鉴意义。本文将逐步均值回归,分位数回归(Lasso Quantile Regression)和弹性网分位数回归(Elastic Net Quantile Regression)三种方法对18只基金进行建模,观察三种方法对风险因子的识别作用,结果发现分位数回归能够很好地对基金收益的尾部进行风格识别,弹性网分位数回归要比LASSO分位数回归包容了更多的风险因子。然后在不同的厌恶参数γ^*情况下,用这三种方法对18只基金的进行了绩效评价,为了判断这些绩效评价的效果,本文设置了三种不同的投资组合方案,在不同的投资组合方案下比较三种方法对基金绩效评价的可靠性,结果表明无论在哪种方案下弹性网分位数回归的评价效果要优于另外两种方法,也间接证明了这种方法识别的风险因子是有效的。Open-end fund is one of the important forms of securities investment business.Their style recognition and performance evaluation are very useful for investors.In this paper,we use stepwise regression,LASSO quantile regression and elastic net quantile regression to model 18 funds,and observe three methods to distinguish the risk factors.The results show that the quantile regression can well identify the tail of the fund’s income,and the elastic net quantile regression contains more risk factors than the LASSO quantile regression.Then using these three methods to evaluate the performance of 18 funds under different aversion parametersγ^*.In order to judge the effect of these performance appraisals,three different investment portfolio schemes are set up in this paper,and the reliability of the fund performance evaluation is compared with the three methods under different portfolios.The results show that the evaluation effect of the elastic net quantile regression is better than the other two methods under any scheme,and it also indirectly proves that the risk factor identified by this method is effective.

关 键 词：开放型基金 弹性网分位数回归 投资风格 基金绩效 

分 类 号：F212.1[经济管理—国民经济] O212[理学—概率论与数理统计]