Journal of Risk Model Validation

Steve Satchell

University of Cambridge

In writing this editorial, I am aware of a backdrop of quantitative easing, which is claimed to bring about increased demand for credit-based products. While my reading on the subject, plus economics 101, seems to suggest that perhaps the main outcome is to weaken domestic currency, and hence allow the country to export more, there are some sound arguments as to why corporate debt, and products based on personal loans, should go up as a consequence of quantitative easing. Thus, it seems convenient that this quarter's issue is devoted to credit/loan issues. Our first paper is by Peter Miu and Bogie Ozdemir, who discuss howto manage capital buffers in the content of Basel II and Basel III. While this is not a precise application of risk model validation, it is sufficiently topical and close to the mark in validation terms to include in our journal.

A more direct application of risk model validation is our second paper, by Joseph Breeden. This paper proposes a direct test for missing terms in retail lending models. In particular, he discusses procedures aimed at detecting the presence of cross-terms in the factor structure. This is very relevant to model failure issues in a number of dimensions. In equity models, for example, we know that many factors, such as value or momentum, behave differently if we sort our factor portfolios by size. These nonlinear factors seem to matter more in periods of financial turbulence.

The third paper, by Sebastian Döhler, considers the problem of validating credit default probabilities. He uses multiple testing procedures to do this. As those of us with strong statistical backgrounds know, this is a challenging problem if we wish to control the type-I error of the testing process. The validity of the proposed process is examined by using empirical and simulated data.

Finally, Meko So and Lyn Thomas look at a way of linking macroeconomic measurements to estimate default risk in credit card portfolios. This is in many ways the “holy grail” problem in retail loan risk management and risk validation, and the authors propose a methodology based on consumers’ scores and Markov chains. Overall, these four papers look at current problems involving loan/account/product risk as a function of macroeconomic/governmental action. As quantitative easing takes us on a macroeconomic roller coaster over the next few years, expect a great deal of new and rich data that will allow us to improve our models in this area.

Our journal is now being indexed and abstracted in:

• Social Sciences Citation Index

• Journal Citation Reports/Social Sciences Edition and

• Current Contents/Social and Behavioral Sciences

This should benefit all parties by giving the journal more recognition and also by making its contents more accessible to new readers.