The Changing Behavior of the European Credit Default Swap Spreads During the Covid-19 Pandemic: A Bayesian Network Analysis

The level of financial risk spread out to the world during the COVID-19 pandemic has shown that none of the countries are immune to financial uncertainty and the vast changes it brings to economic stability. The contagiousness of sovereign risk is a result of the interdependent structure of countries’ financial networks. Yet the analysis of sovereign CDS risk spread using the network view is both new and limited. With this study, we want to use the network view to prove the interconnectedness of the financial systems in Europe and its effect on the spread of the risk throughout the COVID-19 pandemic. The objective of this study is threefold: First, using the Bayesian networks learned from the daily CDS values of 17 European Union countries, we demonstrate the dependent network structure of countries and the movement of the sovereign risk over this network with a cascading behavior. Second, we explore how the probabilistic dependency structure changes over the different phases of the COVID-19 pandemic, leading to alterations on the behavior of the sovereign risk spread. The previous studies on the sovereign risk spread during the COVID-19 pandemic employs the data over the whole period of the pandemic. However, during the pandemic the behavior of the spread was changing, and to capture that change the consideration of shorter intervals becomes crucial. Therefore, in this study, the COVID-19 crisis period from December 2019 until February 2021 is divided into five phases of 3-month time intervals. As the third and last objective, this study intends to be a roadmap for policy makers as well as for researchers to understand the true nature and connectedness of sovereign risk transmissions. For that purpose, we provide a benchmark procedure for the evaluation of the sovereign risk of countries using Bayesian networks which involves a comprehensive analysis involving several steps conducted on each of the learned Bayesian networks for the different phases of the pandemic. In terms of policy implications, this study aims to be helpful for investors that want to diversify sovereign risk in their bond portfolios and be explanatory for the changing behavior of the risk spread during crisis periods. Moreover, this study exemplifies the use of artificial intelligence methods to understand the working mechanism of economic systems.