基于大数据与超网络融合的交易对手信用风险传染模型研究

Considering the interaction between underlying assets (credit and loan, trust, bonds) and derivative assets (credit derivatives), the multi-level credit correlations among counterparties, the multi-dimension of the flow among counterparties, the multi-attribution or multi-criteria of counterparty and so on. Therefore this project integrates the theories and modeling methods of big data and supernetwork to depict complex credit-related features and credit decision-making mechanism among counterparties, to build the supernetwork model of counterparty credit correlations, to dig out the supernetwork structure features of counterparty credit correlation and its internal formation mechanism, and to visualize the formation and correlation of credit risk among counterparties. Based on that theories and methods, it constructs contagion models of counterparty credit risk basing on credit-linked supernetworks, studies the contagion effects of counterparty credit risk caused by single or multiple shocks in the evolution of credit-linked supernetworks, and reveals the mechanism and its essential law of the contagion of counterparty credit risk under credit-linked supernetwork. On the basis of the above studies, using computational experiments and simulation, comparative analyzes dynamic monitoring strategies and its control effects of the contagion of counterparty credit risk. This study can provide effective theoretical and practical guidance for maintaining financial security and raising the level of financial risk prevention.

考虑到标的资产（信贷、信托、债券）与衍生资产（信用衍生品）在交易对手间的交互性、交易对手间信用关联的多层级性、交易对手间流量的多维性，以及交易对手的多属性或多准则等因素的复杂交互作用，本项目融合大数据与超网络的理论和建模方法，刻画交易对手间复杂的信用关联特征及其信用决策机制，构建交易对手信用关联超网络模型，挖掘交易对手间的信用关联超网络结构特征及其内在形成机制，可视化交易对手间信用风险形成与关联特征。在此基础上，构建基于信用关联超网络的交易对手信用风险传染模型，研究信用关联超网络演化过程中单重或多重冲击形成的交易对手信用风险传染效应，揭示信用关联超网络下交易对手信用风险传染机理及其本质规律。基于上述研究，利用计算实验与仿真模拟，对比分析交易对手信用风险传染的动态监控策略及其控制效果。本研究可以为维护金融安全、提升金融风险防范水平提供切实有效的理论依据和实践指导。

本项目通过吸收和运用大数据理论、超网络理论、金融市场投资理论、行为金融学、博弈学习理论、信用风险管理理论、计算实验与仿真模拟等理论与研究方法，融合大数据、复杂系统理论与超网络理论，从企业行为偏好、信用交易行为、流动性传导、投资者情绪、数字技术应用等角度探析了基于信用关联超网络的交易对手信用风险传染影响因素、路径及机制。在此基础上，运用复杂系统理论、超网络建模法与传染病模型等交易对手信用担保网络模型、CDS网络模型、银企多金融关联网络模型、多层耦合网络模型等，探究了不同信用关联网络及其结构下交易对手信用风险传染动力学行为及其演化规律。最后，从流动性救助、舆情控制、事件冲击缓释等角度构建了不同网络结构下交易对手信用风险传染控制模型，并借助仿真分析和对比研究解析了不同策略组合下交易对手信用风险传染控制效果及其演化特征。