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 ›› 2024, Vol. 32 ›› Issue (8): 50-60.doi: 10.16381/j.cnki.issn1003-207x.2022.1429cstr: 32146.14.j.cnki.issn1003-207x.2022.1429

• • 上一篇    下一篇

基于最优异质收益率因子的资产定价研究

倪宣明1,郑田田1,赵慧敏2(),武康平3   

  1. 1.北京大学软件与微电子学院, 北京 100871
    2.中山大学管理学院, 广东 广州 510275
    3.清华大学经济管理学院, 北京 100084
  • 收稿日期:2022-06-30 修回日期:2022-12-18 出版日期:2024-08-25 发布日期:2024-08-29
  • 通讯作者: 赵慧敏 E-mail:zhaohuim@mail.sysu.edu.cn
  • 基金资助:
    广东省自然科学基金项目(2022A1515011893);国家自然科学基金项目(71991474)

Asset Pricing Based on the Optimal Idiosyncratic Return Factor

Xuanming Ni1,Tiantian Zheng1,Huimin Zhao2(),Kangping Wu3   

  1. 1.School of Software and Microelectronics, Peking University, Beijing 100871, China
    2.School of Business, Sun Yat-sen University, Guangzhou 510275, China
    3.School of Economics and Management, Tsinghua University, Beijing 100084, China
  • Received:2022-06-30 Revised:2022-12-18 Online:2024-08-25 Published:2024-08-29
  • Contact: Huimin Zhao E-mail:zhaohuim@mail.sysu.edu.cn

摘要:

本文从经典因子模型的异质收益率出发,通过在残差空间中进行投资组合优化构造异质收益率因子来识别基准因子模型中的遗漏信息,从而对基准模型下的基于异质收益率的资产进行定价,提升基准模型的定价能力,并进一步证明了该拓展因子对异质收益率的资产定价能力。之后,本文基于A股1995年1月—2022年11月的6个因子数据集和美股1963年7月至2022年10月的4个因子数据集,在三因子、四因子、五因子模型的基础上加入异质收益率因子,并将拓展后的模型分别与原模型、均值方差有效单因子模型、主成分分析因子模型的定价效果进行对比。结果显示,在加入异质收益率因子之后,测试资产的α的绝对值均值、GRS统计量和t统计量均大幅减小,原模型的定价效果显著提升,该结果在样本内、样本外,A股、美股市场上都成立,表明了基于异质收益率因子的资产定价的稳健性和适应性。

关键词: 资产定价, 因子模型, 异质收益率, 投资组合优化, 遗漏因子识别

Abstract:

Starting from the idiosyncratic returns of the classical factor model, an idiosyncratic return factor is constructed by optimizing the portfolio in the residual space to identify the missing information in the benchmark model, so as to price the idiosyncratic returns under the benchmark and improve the benchmark. Furthermore, the pricing ability of the extended factor is proved through mathematical derivation. Next, based on 6 factor datasets of A shares from 1995-01 to 2022-11 and 4 factor datasets of the US stocks from 1963-07 to 2022-10, the idiosyncratic return factor is added to three-factor, four-factor, and five-factor models, and the pricing ability of the expanded models is compared with their benchmarks, the mean-variance efficient (MVE) model, and the principal component analysis (PCA) model. The empirical results show that after adding the idiosyncratic return factor, the GRS statistic and t-statistic are greatly reduced, and the pricing ability of the original model is significantly improved, better than the MVE model and the PCA model in most cases. These results hold both in-sample and out-of-sample, for A-share and the US stock markets, indicating the robustness and adaptability of the idiosyncratic return factor.

Key words: asset pricing, factor model, idiosyncratic return, portfolio optimization, missing factor identification

中图分类号: