Credit Risk Contagion Based On Multiagent Endogenous Financial Networks Model

In the current economic system,the correlation among agents has become increasingly close and complex,constituting a huge multiagent and multilayer complex financial network.In this network,due to the highly interconnected of each agent,once a crisis occurs in some agents,it can spread rapidly through the complex network of highly connected agents,and may even spread to the entire network of agents,thus exploding systemic financial risks.Specifically,due to the highly interconnected and complex credit relationship network among agents,once some agents default on credit,risk contagion may occur through the credit network,causing huge losses to the entire economic system.Therefore,it is necessary to study the credit risk contagion among multiple agents in financial networks.In this paper,we summarize the existing studies and construct a multiagent endogenous financial network model including banks,firms,households,government and central bank,and further study the credit risk contagion among the agents based on this model.Firstly,this paper summarizes the concepts related to complex network theory and sorts out the network structure characteristics,such as degree distribution and node weights,to lay the foundation for the subsequent research on credit risk contagion among multiple agents.Further,this paper compares the evolution rules of complex network evolution theory and the application of multiagent modeling theory to provide support for the construction of subsequent multiagent endogenous financial network models.Secondly,this paper sorts out and summarizes the production and operation behaviors of various agents in reality,and then establishes a multiagent behavior mechanism model.After that,the model is simulated and evolved,and the correlations networks among the agents are obtained,including supply chain network,commercial credit network,equity investment network,bond investment network,loan network,savings network and interbank network.Further analysis reveals that in terms of structural characteristics,the network degree distributions are all characterized by power-law distributions,and in terms of size distributions,the size distributions of the banks,firms and households are also characterized by power-law distributions,which is consistent with relevant empirical studies.Meanwhile,the model robustness is verified by changing the key parameters of the model for repeated experiments.Thirdly,based on the behavioral mechanism among agents,the credit risk correlation among agents is sorted out and summarized,and the credit risk contagion mechanism among agents is proposed,and then the credit risk contagion among agents is studied on the basis of the network when the model evolves to a stable state.In the study of the impact of firm nodes on credit risk contagion,the firm nodes with the greatest/minimum degree and the greatest/minimum weight were shocked respectively,and the study found that shocking the firm nodes with the greatest degree and the greatest weight caused large losses in all three sectors,and shocking the firm nodes with the greatest weight also revealed a staggered contagion of risk between firms and banks.In the study of the effect of bank nodes on credit risk contagion,the bank nodes with maximum/minimum degree and maximum/minimum weight are also shocked respectively,and it is found that the credit risk caused by shocking the bank nodes spreads from the banking sector to both the firms and households,and both show staggered bankruptcy peaks for both firms and banks.In the study of the impact of household default on credit risk contagion,the impact of large-scale credit defaults in the household sector is simulated.It is found that in the event of large-scale credit defaults in the household sector,the credit risk caused by such defaults will lead to large losses in the banks,and there will be cross-contagion of risk between banks and firms.Most of the existing studies on credit risk contagion among agents have focused on a single channel of contagion between banks and firms.Based on the existing research,this paper constructs a financial network model including multiple types of agents to further investigate the credit risk contagion among multiple types of agents.The possible innovations of this paper are as follows:(1)The addition of households,government and central bank to the agent model extends the existing research on the construction of multi-agent financial network models;(2)Multiple financial networks are generated based on the behavioral mechanism model of multiple types of agents,and then the credit risk contagion among agents under multiple risk contagion channels is investigated;(3)The model sets the correlation network between agents to be generated endogenously according to the behavior mechanism of each agent.