Models assessing risk
I have just read a interesting article from the Harvard Business Review, most of their articles are “pay per view”, but this was a free one. The most interesting part for me was a small segment on the models used for assessing risk.
Pretty obvious stuff, but some of the comments were:
Models are not decision makers; people are – the issue you have is the culture around modelling.
There are 2 type of risk managers; Quantitive enthusiasts who believe in two types of risks – ones that have been successfully modelled and ones that haven’t. The other type of risk managers are quantitive sceptics who overemphasize the weakness of risk models.
I actually think I sit in the middle – I certainly appreciate the effectiveness, the low cost and high speed delivery that decisioning models provide but also acknowledge that fundamental credit analysis is best performed by professional credit analysts. The question is; have the Banks lost access to these professionals over recent times with the heavier reliance on models.
The pitfall for models is obviously the models reliance on a stable relationship between the input data and the credit risk, often this relationship is determined by past events. Relationships that seem stable during benign conditions may not be so during times of stress. Of course the qualitative nature of risk analysts also have pitfalls, such as the inconsistent views for one analyst to another and the ability to be influenced by emotional (or other) factors.
The key for Banks will be the right blend of well calibrated and consistently validated models with a good level of seasoned professional experts utilising sound judgement.
Related posts
(No Ratings Yet)