改进的DebtRank算法与系统重要性、系统脆弱性研究

Systemic important institutions,systemic vulnerable institutions based on revised DebtRank algorithm

从影响因子、传播方式以及影响程度三个方面改进了DebtRank算法并应用到银行同业拆借市场模型中.研究发现,不发生机构倒闭时,在改进的DebtRank算法下系统的损失等于各个机构对系统损失的贡献之和;不同的初始冲击造成的系统损失、机构损失相互独立.根据机构的系统重要性、系统脆弱性是系统内生决定的假设,给出了DebtRank算法下机构的系统重要性、系统脆弱性权值与次序的定义,证明了一般意义下机构的系统重要性次序、系统脆弱性次序分别由损失矩阵列向量与初始所有者权益向量、损失矩阵行向量与初始所有者权益向量决定.另外,对我国银行数据的研究表明机构的系统重要性与系统脆弱性不一致,而且随时间变化,最后对系统性金融风险的监管提出了建议.

The DebtRank algorithm is improved from influence factor, communication mode and influence degree and applied to the interbank lending market model. It is found that the loss of the system is equal to the contribution of each institution to the loss of the system under the improved DebtRank algorithm without the occurrence of institutional failure. The system losses and institutional losses caused by different initial shocks are independent. According to the vulnerability of the system, the importance of the system decision by endogenous, gives the definition of system importance, vulnerability weights and order, prove the system order, the general order of systemic importance, systemic vulnerability respectively decided by the loss matrix column vector and initial vector, the loss matrix row vector and the initial vector.In addition, the research on interbank lending market in China shows that the systemic importance of the institution is not consistent with the systemic vulnerability, and changes with time. Finally, policy recommendations for systemic risk regulation are put forward.