CMStatistics 2023: Start Registration
View Submission - CFE
A1797
Title: New discrete time affine models to price sovereign credit risk Authors:  Marco Realdon - Brunel University London (United Kingdom) [presenting]
Abstract: Past literature has shown that continuous time affine credit risk pricing models have appealing counterparts in discrete time, namely ARG and VARG0 models based on autoregressive Gamma processes that need no Feller conditions. The aim is to clarify that ARG and VARG0 models are part of a wider family of tractable affine models based on autoregressive Poisson (ARP) processes. The evidence shows that all tested ARP models fit and predict sovereign CDS prices and their volatility very similarly, as long as each ARP model has the same number of factors. Instead, three-factor ARP models better fit and predict the level, but not the volatility, of sovereign CDS prices than two-factor ARP models.