A Research On The Controllability Of Interbank Lending Network Based On The Analysis Of Driver Node Of Complex Network

The financial crisis has made the supervisory authorities realize that microprudential supervision of individual banks cannot achieve financial stability at the macro level.How to establish a sound macroprudential supervision system to prevent systemic risks and maintain the stability of the financial system has become an urgent problem for supervisory authorities to solve.At the same time,the "14th Five-Year Plan" also proposes to improve the modern financial supervision system and improve the financial risk prevention system.Therefore,taking the macroprudential supervision as the starting point,we put forward a research framework for bank network controllability,and use the theoretical methods of complex network controllability to study the controllability of bank lending network from the identification of driver nodes,control centrality and evolution laws,so as to provide a new scientific decision-making basis for the improvement of the macroprudential supervision system.First,based on the bank balance sheet data,we adopt matrix method,entropy optimization and threshold method to construct the bank lending network model.Based on this,we use the method of network controllability to identify the driver nodes in the bank lending network from 2008 to 2017,and analyze the role and influence of the driver node in the bank lending network.The study found that the values of driver nodes were 0.46,0.48,0.32,0.17,0.1,0.14,0.23,0.28,0.33 and 0.39,which was the largest in 2008-2009 and the smallest in 2012.From an economic perspective,the values of driver nodes provide the possibility for the regulatory authorities to implement targeted regulatory policies on the controllability of the bank network.The global financial crisis occurred in2008-2009,and the bank lending network was in a very unstable state.At this time,there were a large proportion of bank nodes to affect the controllability of the whole interbank lending network;in 2012,the controllability of the bank lending network was optimal,with the smallest value of driver nodes,and the central bank needed to monitor fewer driver nodes.On the whole,there is a certain correlation between the value of driver nodes and financial crisis events.When a systemic shock event occurs,the value of driver nodes is often higher.In addition,it is also found that the driver node is not a node with a high degree,nor a node with a large scale.This is a supplement to the previous research results.At the regulatory level,the central bank should consider the overall controllability of the interbank lending network,focusing on relatively small banks as well as large state-owned banks.Sometimes the huge cost to rescue large banks may have little effect,and liquidity rescue to driver bank nodes is a better rescue strategy.The research results add a new dimension from the perspective of network controllability and supervision,identify the driver nodes in the bank lending network,and help supervision determine the institutions that should be supervised.Secondly,we introduce the concept of control centrality for the first time to quantify the control ability of a single bank node over the entire bank lending network,calculate the distribution of node control centrality in bank lending network,focus on the analysis of the control centrality of the driver node,and discuss the relationship between control centrality and degree distribution and layer index.The study found that with the passage of time,the control centrality of driver nodes increased first and then decreased.In2008-2009,the control centrality of driver nodes was relatively small,and the central bank needed to provide greater liquidity assistance to maintain the stability of the lending market.That is to say,the regulatory cost is relatively large;the control centrality of the driver node reached the maximum in 2012,indicating that the controllability of the bank lending network is optimal at this time,and the liquidity rescue cost is the smallest.On the whole,the trend of driver node control centrality is opposite to the trend of its value.The larger the value of the driver node,the more nodes that affect the controllability of the bank lending network.The smaller the control centrality of the node,the greater the monitoring cost of the central bank,the trend of control centrality verifies the correctness and validity of the trend of the value of the driver node.In addition,it is also found that the control centrality of the bank lending network is related to the degree distribution of the network,and the control centrality of the node is related to its layer index.In the bank lending network in the same year,the driver node has a higher layer index and greater control centrality.The research results help the central bank to more accurately understand the control centrality of the nodes in the bank lending network,so as to formulate precise and effective macroprudential supervision policies.Finally,we innovatively study the relationship between the controllability of the bank lending network and the time scale,construct the bank's quarterly,semi-annual and annual lending networks in three time dimensions,examine the changes of the topological properties of the bank lending network and analyze the evolution law of the driver nodes in different time dimensions.The study found that when the average clustering coefficient is maintained at around 0.3,the bank lending network is relatively stable,when it rises above 0.5,the central bank should maintain a vigilant attitude to prevent systemic risks.In addition,it also found that as the time scale increases,the value of the driver node presents a power-law decay law,power-law exponent ? =-0.46,and the ratio of the value of driver nodes is smaller than that of the time scale.The value of driver nodes is different at different levels of time scales,which requires different control strategy.The results show that the regulatory authorities must adjust the policy according to the time scale so that the policy can play a targeted role.In summary,we conduct a comprehensive and systematic study of the controllability of bank lending network.The research results provide quantitative indicators to help regulators develop more effective regulatory and intervention policies,and provide new ideas and basis for improving the macroprudential regulatory framework.