基于鲁棒优化和Copula的ESG投资组合选择研究

As an emerging concept, ESG (Environment, Society and Corporate Governance) investment is increasingly favored by investors. The subjectivity and uncertainty of ESG evaluation affect the quality of investment decisions. How to provide a robust ESG investment strategy is the key and challenge for portfolio selection problem. Therefore, effective decision-making methods are urgently needed to provide decision support for investors. The use of robust optimization to deal with the sensitivity of model parameters is currently the mainstream method in the field of portfolio selection.This proposal firstly models the uncertainty of financial returns and sustainability based on the polyhedral uncertainty set, and constructs a robust mean-VaR portfolio model considering ESG criteria. Secondly, this proposal models the distribution uncertainty of financial return and sustainability return based on time-varying mixture copula model and mixture vine copula methods, and constructs two worst-case CVaR models. Finally, the rationality and effectiveness of the robust models are verified based on the rolling-sample method.This proposal extends the robust optimization theory to the ESG portfolio selection problem and proposes new robust ESG portfolio selection models to provide investors with new robust sustainable investment strategies.

ESG投资作为一种新兴理念日益受到投资者的青睐，而ESG评估的主观性和不确定性影响着投资决策的质量，如何提供稳健的ESG投资策略是目前投资组合选择问题所要解决的关键和挑战。因此，迫切需要有效的决策方法来为投资者提供决策支持。运用鲁棒优化思想处理模型参数的敏感性问题是目前投资组合领域的主流方法。本项目首先基于多面体不确定集对财务收益与可持续性的参数不确定性进行建模，构建考虑ESG准则的鲁棒均值-VaR投资组合模型；其次，分别基于时变混合copula和混合vine copula方法对财务收益与可持续性收益的联合分布不确定性进行建模，构建WCVaR模型；最后，基于滚动样本方法验证所构建鲁棒模型的合理性与有效性。本项目将鲁棒优化理论扩展到ESG投资组合选择问题中，提出新的鲁棒ESG投资组合选择模型，为投资者提供新的稳健的可持续投资策略。

近几年，ESG投资理念日益受到市场的重视。如何构建稳健的ESG投资策略具有重要的现实意义。按照项目研究计划，本课题主要从三个方面开展了研究工作。首先，在传统的投资组合框架中引入ESG约束，使用多面体不确定集捕捉资产的财务回报和ESG表现的不确定性，构建了参数不确定下的鲁棒ESG投资组合选择模型。其次，绿色债券是满足ESG准则的投资工具之一。在均值-CVaR和WCVaR框架下，基于动态R藤Copula模型分析了绿色债券的投资价值。最后，基于机制转换R藤copula模型和非对称 GARCH-MIDAS 模型分析了复杂多变环境下的动态投资策略。本课题的研究成果为可持续投资决策管理提供了新的工具。

内容获取失败，请点击重试