Misplaced reliance on sophisticated maths?
Posted on 31-03-2009 | 0 comments
The UK regulator’s Turner report and comment in the financial press (Financial Times “Maths and markets” editorial, March 21) succeed in highlighting the important role of financial models to the financial services industry. Lessons drawn in the context of the banking sector fiasco are likely to be applicable elsewhere so the issues raised are worthy of careful consideration. However, as a self-confessed geek, in the business of bridging the gap between clever theory and pragmatic application, I believe commentators fail to give sufficient weight to two major overarching challenges in the use of models.
The basic task of the modeller is two-fold. First, for a specific task, to select and implement a suitable mathematical model from today’s vast array of possibilities. It is true that these models can (and will) always be improved but, believe me, there is already plenty of choice. Second, to choose assumptions i.e. to ‘calibrate’ the model to some window of past data or some other source of information or expert opinion. At best models can give valuable management information and insight into risk exposures. At worst they can obscure reality and create a false sense of security. The potential for poor choices gives rise to what is known as ‘model risk’. In some situations these model risks may be ‘first order’ (i.e. they really matter) in decision making and so the modeller needs to be much more than a mathematician.
However, in my opinion, by far the most difficult (and potentially costly) challenges for modellers arise in their interactions with firm management, accountants and regulators. First, the more sophisticated the maths, the less likely are senior management to understand models and their sensitivities and vulnerabilities. Does the modeller simplify (and obscure) or build more complexity (and obscure)? It does suggest that firms who choose to hold complex exposures need strong technicians at a very senior level. Second, is the apparent lack of incentives for senior management to fully incorporate the available insights of the models into their business model and decision making. A strong governance framework can provide some protection from the risks which flow from choice around models and assumptions. Too often these choices have been left unchecked by the lack of deep knowledge and credulity of regulators and accountants. Meeting the challenges posed does not just require smarter maths but also a focus on governance, managerial technical competence and understanding model risk. Whilst all stakeholders would tell us these questions are on their respective agendas, we all now need to give real force to the words.
Please read our participation policy before submitting a comment.