文本描述
基于极端约束的投资组合选择及有效性研究
摘
要
近年来,国际形势发生深刻变化,新冠肺炎疫情在全球持续蔓延。在此背景下,世
界多国资本市场屡遭重创,中国股市也难以独善其身。此外,随着外部环境的变化和国
内经济供给侧结构性改革的深化,中国经济增长也面临着一些新的困难和挑战,金融市
场难免会受到多种风险因素的冲击,极端风险事件引发股市剧烈波动的可能性不容忽视。
因此,从未雨绸缪的角度考虑,国内机构和个人投资者将极端约束纳入投资决策过程当
中,建立有效应对极端风险的投资组合,具有重要的理论和现实意义。
本文以现代投资理论和方法为基础,将极端约束条件纳入投资者的投资组合选择过
程,系统深入地研究基于极端约束的投资组合选择机理及其背后所反映的应对“极端市
场风险”的投资行为。极端约束包含极端损失约束和极端敏感性约束。极端损失约束是
指当资本市场受到极端风险事件冲击时,投资者要求其持有的证券组合的最大损失率不
超过某一给定水平;极端敏感性约束是指在极端不利情况下,投资者要求控制其所持有
的证券组合相对于市场组合波动的敏感性。传统均值 -方差模型由于没有考虑资产收益
率非正态分布特征,从而会低估有效组合的尾部风险;均值-TEV模型主要关注投资组
合的跟踪误差方差大小,而忽视了对投资组合总体风险的控制。基于此,本文分别建立
并求解了基于极端约束的均值-方差模型和均值-TEV模型,运用沪深 300十大行业指数
从 2005年 1月到 2020年 8月的月收益率数据进行数值分析,并实证检验不同极端约束
条件下的投资组合策略的有效性。
首先,在传统均值-方差投资组合选择模型中纳入极端损失约束和极端敏感性约束
条件,分别建立了基于极端损失约束和基于极端敏感性约束的均值-方差模型,并通过模
型求解和数值分析,分别考察了极端损失约束和极端敏感性约束条件下的均值 -方差投
资组合选择。理论分析结果表明,基于极端损失约束的有效组合满足 K+2基金分离定理
(其中 K为极端损失约束中的有效紧约束的数量);数值分析表明,在投资组合选择过
程中施加极端损失约束条件,不仅能够改善有效组合收益率的分布,而且还有利于提升
有效组合的收益率偏度和上方获利能力。基于极端敏感性约束的均值 -方差有效组合满
足 3基金分离定理;极端敏感性约束对有效边界的影响与组合期望收益率水平和极端敏
感性程度有关;当组合期望收益率相对较低且极端敏感性大小适中时,在投资组合选择
过程中施加极端敏感性约束条件,能够提升有效组合的收益率偏度。
其次,在最小跟踪误差方差(TEV)投资组合选择模型中纳入极端损失约束和极端
敏感性约束条件,分别建立了基于极端损失约束和基于极端敏感性约束的均值-TEV模
型,通过模型求解和数值分析,分别考察了极端损失约束和极端敏感性约束条件下的均
值-TEV投资组合选择。结果表明,基于极端损失约束的均值-TEV投资组合满足K+3基
I
摘
要
金分离定理,在均值-TEV模型中纳入极端损失约束,能够提高投资组合的下方风险控
制能力和上方获利能力,从而有利于提高投资组合收益率的偏度。基于极端敏感性约束
的均值-TEV投资组合满足 4基金分离定理;进一步研究发现,极端敏感性约束对均值-
TEV有效边界的影响与组合期望收益率水平和极端敏感性大小有关,当组合期望收益率
相对较低且极端风险程度适中时,极端敏感性约束有利于提升均值-TEV组合收益率的
偏度。
最后,以最小方差组合、跟踪误差方差最小化组合为例,运用国内 A股市场的数据
实证考察了基于极端损失约束和极端敏感性约束的投资组合策略的样本外业绩,并与等
权组合和传统最小方差组合、最小跟踪误差方差组合的业绩进行比较。实证结果表明,
在传统最小方差组合和最小跟踪误差方差组合中加入极端损失约束和极端敏感性约束,
不仅能够提高组合的样本外的平均收益率和夏普比率,而且在考虑交易成本的情况下也
能带来比等权组合和传统投资组合更高的净夏普比率。相对于极端损失约束,极端敏感
性约束更能提升组合的夏普比率。而且,基于极端损失约束和极端敏感性约束的投资组
合策略的有效性相对于样本数据选择区间具有稳健性。
以上研究结论不仅有助于理解投资者的投资行为,同时也为投资者选择合适的投资
策略以应对极端风险事件的冲击提供证据支持。
关键词:极端损失约束;极端敏感性约束;投资组合选择;投资策略有效性
II
基于极端约束的投资组合选择及有效性研究
Abstract
In recent years, profound changes have taken place in the international situation, and the
COVID-19 continues to spread around the world. Against this background, the capital markets
of many countries in the world have been hit hard repeatedly, and China's stock market is hardly
immune to it. In addition, with the change of the external environment and the deepening of the
supply side structural reform of the domestic economy, China's economic growth is also facing
some new difficulties and challenges. The financial market will inevitably be impacted by a
variety of risk factors. The possibility of extreme risk events causing sharp fluctuations in the
stock market cannot be ignored. Therefore, from the perspective of never taking precautions, it
is of great theoretical and practical significance for domestic institutions and individual
investors to incorporate extreme constraints into the investment decision-making process and
establish an effective portfolio to deal with extreme risks.
Based on modern investment theories and methods, this paper brings extreme constraints
into the process of investors' portfolio selection, and systematically and deeply studies the
mechanism of portfolio selection with extreme constraints and the investment behavior behind
it to deal with "extreme market risk". Extreme constraints include extreme loss constraint and
extreme sensitivity constraint. Extreme loss constraint refers to that when the capital market is
impacted by extreme risk events, investors require that the maximum loss rate of the securities
portfolio held by them not exceed a given level. Extreme sensitivity constraint refers to that
under extreme adverse circumstances,investors require to control the sensitivity ofthe
securities portfolio they hold to the volatility of the market portfolio. The traditional mean-
variance model underestimates the tail risk of effective portfolio because it does not take into
account the non-normal distribution characteristics of asset returns. The mean-TEV model
mainly focuses on the tracking error variance of the portfolio, while neglecting the control of
the overall risk of the portfolio. Based on this, this paper establishes and solves the mean-
variance model and mean-TEV model with extreme constraints respectively, uses the monthly
yield data of the Shanghai Shenzhen CSI 300 Index of China stock markets from January 2005
to August 2020 for numerical analysis, and empirically tests the effectiveness of portfolio
strategies under different extreme constraints.
First of all, the traditional mean variance portfolio selection model includes extreme loss
constraint and extreme sensitivity constraint. The mean-variance model with extreme loss
constraint and extreme sensitivity constraint are established respectively. Through model
III