A spatial analysis of contagion in sovereign credit default swaps

In this article, we propose a spatio-temporal model to investigate the dynamics of contagion in the credit event risks of sovereigns. More specifically, we examine how changes in the credit default swap (CDS) spreads of a sovereign are influenced by the CDS spreads of other sovereigns over time. Our model incorporates spatial, temporal, and spatio-temporal lags of CDS spreads while accounting for unobserved heterogeneity across sovereigns and time periods. We consider several candidates for the underlying contagion network matrix using cross-border domestic bank exposures, geographical distances between sovereigns, and pairwise correlations of CDS spreads. We propose an efficient Bayesian algorithm for estimation and a simple method to address nested and non-nested model selection problems. Using a quarterly dataset of fourteen sovereigns from 2009 to 2022, we find evidence of contagion in CDS spreads which was relatively stronger during the period 2009–2012.