Abstract

Purpose: Within literature as well as in practice there is no one common standard of
measuring portfolio credit risk of credit exposures toward retail clients. Therefore,
the Article’s goal is to present the possibility of using for these purposes so called
factor models.

Approach: The article presents theoretical analysis of factor models’ properties and
then is illustrating theirs usage basing on the data on mortgage backed housing loans
and cash loans from one of the biggest credit institutions licensed in Poland. In the
article techniques of stress testing analysis are being used as well.

Findings: The significant differences in credit value at risk estimates indicate that
parameters and assumptions of factor models have a critical role when we try to use
this group of models. Additionally, the results suggest that it is necessary to use verification
tools like stress tests or expert knowledge.

Value: The article extends the literature findings in two aspects. First, it presents
how to use factor models for retail loans portfolios and shows the results of CVaR
estimates basing on empirical data. Second, it verifies sensitivity of CVaR estimates
to model selection and risk parameters.

Keywords