安徽工程大学学报2025，Vol.40Issue(3)：87-94,8.
ESG对强化学习的连续时间均值-方差投资组合的影响
The Effect of ESG on Continuous Time Mean-variance Portfolio Based on Reinforcement Learning
摘要
Abstract
In the context of"double carbon"goal,ESG investment concept has gained wide attention.The enterprise ESG rating and investor ESG preference are incorporated into the continuous time mean-variance portfolio model under the framework of reinforcement learning to achieve the optimal trade-offbetween exploration and utilization.Secondly,the portfolio problem is transformed into a relaxed stochastic control problem with entropy regularization,and it is proved that the optimal feedback strategy has a Gaussian distribution and the variance decays with time.Next,the Hamilton-Jacobi-Behrman(HJB)equation satisfying the value function of the self-financing portfolio problem is derived,and the strategy improvement theorem is further proved,and the RL algorithm is designed for different ESG rating levels.Finally,the behavior characteristics of investors'portfolio decision are analyzed by numerical simulation.The results show that focusing on ESG factors leads to investment opportunities,optimized decisions,and better returns for portfolios with high ESG ratings.关键词
强化学习/熵正则化/ESG标准/均值-方差投资组合/哈密尔顿-雅可比-贝尔曼方程Key words
reinforcement learning/entropy regularization/environment,society,governance(ESG)/mean-variance portfolio/Hamilton-Jacobi-Bellman(HJB)equation分类
管理科学引用本文复制引用
贵秀子,费为银,邓寿年..ESG对强化学习的连续时间均值-方差投资组合的影响[J].安徽工程大学学报,2025,40(3):87-94,8.基金项目
国家自然科学基金项目(62273003) （62273003）
安徽未来技术研究院企业合作项目(2023qyhz07) （2023qyhz07）
