Credit Risk Consulting
Insights, Modelling, Solutions, Guidance
Research
In the last few years lenders have had to stretch their analytical capabilities in constructing models for the Basel Accord. We have seen some analytical problems that many lenders have faced but which have never been properly resolved, leading to potentially inappropriate modelling solutions.
With the rush for I.R.B. compliance mostly over, we believe that the time is right to address these issues. Typically, research projects could be (a) run by ourselves, or (b) we could act as consultants to internal teams or (c) we could work with key university experts in the case of difficult topics.
Some starter possibilities are:
- Examine the use of Mixed Effects Models for situations where there are multiple observations for the same account. A correct statistical approach in this area would improve behavioural modelling (e.g. behaviour scorecard developments)
- Examine the use of Hierarchical Bayesian techniques for low-default-portfolio modelling. This has particular relevance for corporate lending
- Provide a best-practice guide with example analyses for the sorts of econometric models that have been used for Long-Run PD calculations
We welcome discussion on these and other areas.
Avenir Risk Limited