Title: Modelling spatial contagion effects for mortgage defaults using a Bayesian hierarchical approach
Authors: Matteo Spada - Paul Scherrer Institute (Switzerland) [presenting]
Raffaella Calabrese - University of Edinburgh (United Kingdom)
Abstract: The aim is to explore how the spatial location of US mortgage loans can improve the predictive accuracy of credit risk models. The recent financial crisis led to much concern about the so-called credit contagion - how the deterioration of a borrowers future ability to honour its debt obligations can affect the ability of other borrowers living in the same neighbourhood. Consequently, the spatial contagion effects is suggested to be included in the risk assessment for mortgage loans. A Bayesian hierarchical model based on the geographical locations of properties is proposed to measure the impact of the credit contagion on the overall risk. This is a robust and fully probabilistic approach that incorporates spatial correlation and enables to model both the observed data and any unknowns as random variables. Such a method is applied to loan-level data released by Freddie Mac on US single-family mortgages.