Title: Dynamic default intensities in a network topology
Authors: Xiu Xu - Mrs (Germany) [presenting]
Abstract: Systemically important banks are connected and have dynamic dependencies of their default intensities. Here a network view on these dependencies is developed based on extending the Dynamic Nelson Siegel model. Using connectedness measures built from variance decomposition, one can study the dynamics of these default intensity curves in terms of network evolution. After extracting the underlying long-run, short-run, and median-run networks from credit default swap spreads among different maturities from 2008-2015,we find that: 1. The total connectedness of default risk among maturities is evidently characterized: long-run > short-run > middle-run; 2. It presents unprecedented directional connectedness surge from Europe to US since 2012, especially during the European sovereign debt crisis periods, behind the significant spill-over of directional connectedness from US to Europe during 2008-2011. Additionally, the diverse drivers of long-run, short-run,and middle-run connectedness of default intensity are thoroughly analysed. Moreover,the network DNS model reveals favourable outperformance in out-of-sample forecasting CDS spreads, especially for European banks at sovereign debt crisis periods.