Core-middle-periphery network model for China banking system

Abstract

The banking system could be mapped by the network model to generate the structural properties of evolution dynamics. In this study, we empirically investigate the evolution properties of the China bank network from 2008 to 2019 where the banks and lending relationships are set as the nodes and links. By introducing the middle layer into the core–periphery (CP) model, we present the core–middle–periphery (CMP) model where the nodes belonging to the core layer are fully connected and the ones belonging to the middle layer connect the core and periphery layer. Compared with the traditional CP model, the reconstruction error of the CMP model is decreased by 64% compared with the one obtained by the CP model, and the transition stability probability is enhanced greatly. This work is helpful for deeply understanding the evolution properties of the banking system.