文章摘要
Ma Jingyi,Wang Zi.Combined Portfolios from Sequentially Non-negative Minimum Variance Model and Its Application[J].The Journal of quantitative and technical economics,2021,(6):166-180
社会核算矩阵平衡方法研究
Combined Portfolios from Sequentially Non-negative Minimum Variance Model and Its Application
  
DOI：
中文关键词: 社会核算矩阵  最小二乘交叉熵法  加权平均
英文关键词: Short-selling Constraint  Minimum-variance  Combined Portfolio
基金项目:本文获得国家自然科学基金青年基金项目（71503092）、国家自然科学基金国际(地区)合作与交流项目（71461010701）、中央高校基本科研业务费专项资金资助项目（项目批准号2662020JGPY009）的资助。
Author NameAffiliation
Ma Jingyi School of Statistics and Mathematics, Central University of Finance and Economics 
Wang Zi School of Statistics and Mathematics, Central University of Finance and Economics 
Hits:
Download times:
中文摘要:
      研究目标：探讨在二元信息基础上引入第三元信息时提升SAM表平衡方法精度的边界条件，考察综合利用流量、系数矩阵信息时如何选择加权平均形式以提升平衡方法的精度。研究方法：基于蒙特卡罗模拟思想，在SAM表中引入均匀分布和正态分布随机误差项，通过均方根误差指标来比较平衡方法的有效性。研究发现：在平衡SAM表时，如果流量信息精度较高，则应采取加权算术平均的形式构建目标函数；如果行系数或列系数矩阵精度较高，则应该采取加权几何平均的形式构建目标函数。在额外信息精度较高的情形下，引入额外信息将有助于提升平衡后SAM表的精度。在额外信息精度较差的情形下，当在列系数和行系数矩阵信息基础上以加权几何平均形式引入精度较差的流量信息时，可提高平衡后SAM表的精度；在其余情形下，额外引入精度较差的信息并不会显著提高平衡后SAM表的精度。研究创新：探讨了在目标函数中引入额外信息时提升平衡方法精度的边界条件，并比较了加权平均形式选择对估计精度的影响。研究价值：本研究对于提升SAM表的精度具有重要参考价值。
英文摘要:
      Research Objectives: This paper improves the non-negatively global minimum-variance model and proposes the sequentially non-negative minimum-variance model. Research Methods: We establish the asymptotic properties of the estimates for the sequentially non-negative minimum variance model and then analysis the properties of the expected return and variance of the combined portfolio from the sequential model. These properties are verified through simulation and the history data analysis. Research Findings: Compared with the non-negatively global minimum-variance portfolio, the proposed combined portfolio has lower out-of-sample risk and higher return. In the meanwhile, the proposed method also has an advantage of generating a sparse portfolio. Research Innovations: Under the no-short sell constraint, this paper proposes a novel combined portfolio by sequentially non-negative minimum-variance models. Research Value: The proposed method shows strong applicability in Chinese stock market and this work enriches the current research on modern portfolio theory.
View Full Text   View/Add Comment  Download reader
Close