An empirical study on the correlation structure of credit spreads based on the dynamic and pair copula functions

Abstract

Purpose – The purpose of this paper is to study the correlation structure of the credit spreads. Design/methodology/approach – The minimal spanning tree is used to find the risk center node and the basic correlation structure of the credit spreads. The dynamic copula and pair copula models are applied to capture the dynamic and non-linear correlation structure. Findings – The authors take the enterprise bond with trading data from January 2013 to December 2013 as the research sample. The empirical study of minimum spanning tree shows that the credit risk of corporate bonds forms a network structure with a center node. Meanwhile, the correlation between credit spreads shows dynamic characteristics. Under the framework of dynamic copula, the lower tail dependence is less than the upper tail dependence, thus, in economic boom period, the dynamic correlation is more significant than in recession period. The authors also find that the centrality of credit risk network is not significant according to the pair copula and Granger causality test. The empirical study shows that the goodness-of-fit of D vine is superior to Canonical vine, and the Granger causality test additionally proves that the center node has influence on few other nodes in the risk network, thus the center node captured by the minimum spanning tree is a weak center node, and this characteristic of credit risk network indicates that the risk network of credit spreads is generated mostly by the external shocks rather than the internal risk contagion. Originality/value – This paper provides new ideas for investors and researchers to analyze the credit risk correlation or contagion.