内生性风险可控视角下信用债CDS违约风险缓释效用目标跟踪实现的鲁棒优化研究

Since the bond CDS issued, there still exist some shortcomings about the research on tracking the default risk mitigation utility goal and mastering the endogenous risk, such as, the generated factors of endogenous risks are not mined fully, where the endogenous risks are originated from three counterparties of reference entity, credit protection buyer and credit protection seller; the mathematical structure of the default risk mitigation utility does not achieve the convincing results; the robust optimization method for getting the utility goal of default risk mitigation and mastering the endogenous risks, is still not effective. All the above shortcomings will weaken the ability of the bond CDS. Therefore, we plan to do the following research. By mining deeply the endogenous risk factors generated by counterparties, we explore the mechanism and form the mathematical structure of the mitigation utility. At the same time, we also get the controlling variables set of the endogenous risk and corresponding admissible conditions. Using the merits of the anti-uncertain perturbation, robust and VaR constraints, we try to find a robust optimization method to track the goal of the static or dynamic default risk mitigation utility, as well as controlling robustly the endogenous risks under the worst situation. Based on this, we finally derive the optimal design scheme for the risk control variables, and provide a technical scheme for the design and operation of the China's credit bond CDS.

自信用债CDS面世以来，关于其违约风险缓释效用目标实现和内生性风险控制的研究尚存在如下不足：信用债CDS运转链条上各交易对手（参考实体、信用保护卖方和买方）的内生性风险生成因子，尚未得到充分挖掘；反映其违约风险缓释效用的数学结构，未取得令人信服的成果；既能跟踪实现违约风险缓释效用目标、又能从容驾驭内生性风险的鲁棒优化方法研究，仍不充分。上述不足很大程度上削弱了信用债CDS的质量和生命力。鉴于此，本项目拟进行下列研究：从挖掘各交易对手的内生性风险生成因子入手，探寻其作用机理并构建违约风险缓释效用数学结构，同时析出内生性风险的各控制变量，形成体现内生性风险可控性的控制变量容许集；利用鲁棒抗干扰特质，通过施以鲁棒与VaR约束，探究最坏情境下、跟踪实现静态和动态违约风险缓释效用目标、且稳健驾驭内生性风险的鲁棒优化方法，解析出各内生性风险控制变量的最优设置方案，为我国信用债CDS运作提供技术支持。

随着我国信用债违约事件频发，信用债“刚兑”预期被打破。为应对急剧上升的违约风险敞口，投资者急需有效的金融工具以对冲信用债违约风险，CDS（信用违约互换）对参考实体的违约风险缓释效用强大，中国版CDS（CRMW——信用缓释凭证）应运而生。然而，如何把握CDS对违约主体的违约风险缓释效用亟需解决。. 本项目从挖掘CDS运转链条上各交易对手引发内生性风险的生成因子为突破口，探究跟踪合理的违约风险缓释效用目标、保证实现最低缓释效用、同时将内生性风险掌控在可控范围内的鲁棒优化方法，为我国信用债CDS运作提供技术支持。主要取得了如下重要结果：（1）构建了信用债参考主体的违约风险缓释效用的数学结构，并借助于CVaR思想创建出信用债CDS违约风险缓释效用度量结构；（2）基于管理者不同的跟踪目标需求，分别以市场中所有CRMW平均风险缓释效用、平均总体风险缓释及动态风险缓释三种目标，构建了相应的目标跟踪的债券投资组合优化模型，并通过优化方法分别给出了优化策略；（3）构建了具有CDS保障的信用债投资组合收益模型，采用仿真实验技术得到了最优鲁棒优化策略；（4）构建了SSA——CatBoost可解释机器学习方法，并应用到客户信贷内生性风险因素的挖掘与提取中；（5）构建出PSO-SVM 供应链金融预警模型分析公司信用违约风险；（6）利用UTADIS方法构建出公司信用风险水平判别技术。. 项目在执行过程中主要以债券利率市场数据与CRMW数据为基础展开模拟与实证分析，获得了较好的预期效果。项目所获得的成果对我国利用CDS这一工具对企业债进行风险管理提供了技术参考。

内容获取失败，请点击重试