This study represents an attempt to review, and analytically evaluate, each of the major processes used by industry participants to measure so-called “credit risk” for first mortgage products. The study has three sections: Section I provides a discussion of the general concept of Economic Capital (“EC”), how EC is measured and used by best-practice banks, and how EC concepts used by industry practitioners differ from regulatory definitions of capital.Section II discusses the various types of theoretical “credit risk models” that are used by practitioners to measure EC for mortgages. Section III conducts several empirical experiments in which large historical databases are used to estimate the credit risk models described in Section II.
The empirical work is aimed at helping the practitioner and the regulator to evaluate the results ofalternative models. The study was funded by Washington Mutual, but was conducted by an independent group of researchers whose views are expressed in thispaper.
II. Theoretical Discussion Models for Estimating Unexpected Losses for Mortgages. There are basically four classes of model examined in this paper – 1) so-called Panel Data models, 2) the so-called First Generation Default Mode (FGDM) model, 3) so-called Merton models, and 4) so-called Roll Rate (RR) or, equivalently, State Transition models.
Download
Best-Practices in Mortgage Default Risk Measurement and Economic Capital
