基于CVaR和改进熵的全贷款组合优化模型

贷款组合的配置优化是商业银行资产管理的核心问题。针对该问题，从全部贷款组合整体风险和风险分散度两个角度，构建了基于CVaR和改进熵的全贷款组合优化模型。通过在比例熵中引入调整系数构建改进熵来衡量全部贷款组合的分散化程度，改进了比例熵约束必须给予每个企业一定贷款权重而造成过度分散化的弊端，使得全部贷款组合分散化的衡量更加合理。通过控制存量组合和增量组合叠加后的全部组合风险的情况下建立多目标规划模型，得到增量资产的最优配置。改变了流行研究仅仅控制增量风险的不足。在贷款收益率服从Laplace分布假定的基础上对全部贷款组合进行风险控制，符合金融资产收益率“尖峰肥尾”的特性，使得风险可以准确评估，改进了以正态分布为基础进行研究，不能准确拟合贷款收益率实际分布，进而无法合理估计风险的弊端。

Loan Portfolio Optimization Model Based on CVaR and Improved Entropy

The optimal allocation of loan portfolio is the core problem of commercial bank asset management. To solve this problem, a whole loan portfolio optimization model based on CVaR and improved entropy is constructed from the perspectives of the overall risk and the risk dispersion of all loan portfolios. This paper is innovative because it has measured the degree of diversification of all loan portfolios by introducing adjustment coefficient into the proportionality entropies to improve the entropy, which improves the drawbacks of the excessive diversification caused by the proportion of entropy constraint that must give every enterprise loan weight and makes the diversification measurement of the entire loan portfolio more reasonable. Besides, it has established a multi-objective programming model by controlling the total portfolio risk after the combination of stock portfolio and incremental portfolio to get the optimal allocation of incremental assets. It has changed the popularity of research by only controlling the lack of incremental risk, which changes the lack of only controlling the incremental risk in the popular research. Furthermore, the risk control of all loan portfolio is based on the loan return rates, following the Laplace distribution assumption, which is in line with the “rush fat tails” feature of the return on financial assets, making it possible for the risk can accurately assessed. It therefore, improves the research based on the normal distribution assumption which cannot accurately fit the actual distribution of the loan rate of return, and thus cannot reasonably estimate the disadvantages of the risk.