Risk transmission between banks based on time-varying state network

Abstract

Aiming at the state transition between bank networks, we propose a time-varying state network model. In this model, we classify the bank networks in each time period by the kmeans method, and use directed minimum spanning tree(DMST) method to describe the topological structure of each kind of bank network. We also construct a time-varying bank state network by the planar maximally filtered graph(PMFG) method. The state network can be used to find the source of bank risk and conduct the time-varying analysis. We put into the model the inter-bank lending data of 15 listed Chinese commercial banks from the fourth quarter of 2007 to the first quarter of 2019. The results show that the short-term continuity jump between the bank states can effectively describe the occurrence of financial crisis. For example, before and after the global financial crisis in 2008, there was a short-term jump between two states. From the “money shortage” in 2013 to the stock market crash in 2015, there were four short-term jumps between states. At the same time, the outgoing degree of each directed bank state network is directly proportional to the contagion effect, and the incoming degree is inversely proportional to the steady degree of the risk faced by the bank. The sequential bank state network has the memory characteristic, which can provide the central bank for decision basis to prevent the systematic risk.