On the rating and pricing of mortgage portfolios through structured finance

Kaj Nyström

Abstract

ABSTRACT

During the summer of 2007 the financial markets of the world were rippled by the crisis in the US market for subprime residential mortgage loans. The risk in the subprime residential mortgage loans has been packaged and sold over the world using structured finance instruments such as collateralized debt obligations (CDOs), and as many of these products have been assignedthe highest rating by the rating agencies it is unavoidable to question the methodology used by the rating agencies to assign labels of credit quality to these structures as well as the role(s) played by the rating agencies in theprocess. In this paper we introduce a structured approach to concepts of redit quality and we discuss in detail how to assign a notion of rating to cashflows exposed to credit risk. We claim that the model risk and complexity of the process of pricing non-traded credit risk using structured finance products like cashflow CDOs must be considered as significant and that it seems clear that transparency concerning methods, models, assumptions and definitions of credit quality is crucial to the process of creating an even more efficient market for structured finance products, such as cashflow CDOs. One of our points is that it should be possible to create universal, standardized, transparent and mathematically formalized approaches to the rating of structured finance products. Our discussion is exemplified by a cashflow CDO constructed using a portfolio of residential mortgages as reference portfolio and a key problem is to choose the set of scenarios, to base the analysis of the CDO on, and to attach probabilities to these scenarios. Different sets of scenarios and different sets of probabilities give rise to different cashflows and hence to different risks, credit qualities and prices for the instruments involved. Our example emphasizes the effects of the business cycle and the importance of the timing of defaults and the effect, on losses and risks, of different models as well as the impact of structural differences on the generated cashflows.