The network structure of the China bond market: Characteristics and explanations from trading factors

Abstract

Based on the transaction records of financial institutions in the China interbank bond market, we construct a binary and directed trading network for each trading day. Topology analysis of network snapshots shows the network to be scale-free and disassortative, revealing a stable “core–peripheral” hierarchical structure. Further, the rich-core method is adopted to identify the core nodes and divide the overall network into three subnetworks. Relying on the exponential random graph model (ERGM), we find that the pledge rate is vital in shaping the network structure. Still, collateral has a relatively weak impact as compensation for default risk. Appropriate counterparties, including primary dealers, trading institutions sharing more common counterparties and institutions that have formed long-term trading relationships, are conducive to forming a network. We also find that the transaction cost factor has the least impact on the core subnetwork, and the counterparty factor is more important for the intermediate and peripheral subnetworks.