Assessment of mortgage default risk via Bayesian reliability models

摘要

In this paper, we consider duration-type models and their generalizations for modeling default risk. The models are motivated by noting similarities between reliability/survival analysis and mortgage default risk. We present Bayesian modeling strategies used in reliability analysis for describing time to default data. Our models include proportional hazards-type generalized gamma and mixture models, which are capable of capturing nonmonotonic default rates. We develop Bayesian inference for our models and illustrate their implementation using actual time to default data from the U.S. mortgage market. Copyright © 2010 John Wiley & Sons, Ltd.