Stress testing of retail mortgages: a study based on non-stationary Markov chains and t-copula simulation

Chang Liu, Min Guo and Raja Nassar

Abstract

ABSTRACT

In this study, real-life retail mortgage loan data from a Chinese national commercial bank is used to generate the projected distribution over a set of predefined mortgage states or categories. Non-stationary Markov chain transition probabilities between these states are calculated using loan data from 57 consecutive months. In order to validate the model and assess the risk, we used the data gathered to simulate, by a t-copula method, the projected portfolio distribution over the states of a retail mortgage loan in different shock scenarios. The approach proposed in this paper can be readily extended to other retail credit products as well.