Stress testing and VaR models - finding the wood from the trees

Hi,

http://www.ft.com/reports/risk-management-april2009

Business is hotting up for us and I wasn't going to post anything until next week but the risk management report in the FT and the recent fretting about the Wall St stress tests has left me with some opinions I thought you'd find interesting.

VaR models come in for much criticism in the FT report but the pundits who are commenting on this are missing some basic issues and are still unable to see the wood from the trees. I've mentioned a number of times in recent blogs about the importance of understanding your risk models and their weaknesses. VaR continues to come in for much flak, especially the monte-carlo based models, however, the answer here isn't to throw them out and start building again, it is to adapt your models to the new environment, and make them 'future market proof'.

The scope of how to do this is too large for these pages but I stress, building again from scratch isn't needed if you have operational capacity to produce monte-carlo based VaR numbers. If you understand your model, you can easily adapt it to give you a much better guide on potential losses in future market scenarios. Now isn't the time to invest large sums of money in the alleged latest way to model financial risk. Now is the time for sound 'what if' analysis from your risk management department coupled with a smart hedging strategy.

This leads me neatly in to the Wall St stress tests which is the mother of all 'what if' projects. These numbers will undoubtedly influence share prices, but I suspect they will merely confirm what the market has decided in the past year. I don't think there will be a huge divergence in perceived health beyond that which has been implied already in the last two years of share price moves.

Pre-crisis, stress testing had been too lax with risk managers not able to make the conceptual leap from nonvolatile markets into possible future volatile ones. This is a classic human context mistake. If you are in a benign environment, it is very difficult for you to make the jump into a potential crisis enironment and come up with scenarios based off what a major increase in volatility would look like. Now that volatility has come, I'm sure that some of the possible future scenarios being suggested are the other way - much too conservative and apocalyptic. Despite the fact you may solve the context problem, assessing worst case scenario loss in chaotic markets and coming up with anything like meaningful numbers is an impossible job to do accurately. Stress scenarios are typically driven by liquidity issues and forced unwinds of a defaulted firms positions than assets jumping to certain fundamental levels which banks were able to anticipate in their stress analysis.

As a result, financial firms need to subtly change how they use this information to help to protect themselves in a downturn. They should consider dampening losses through a clever hedging strategy as any possible scenario, however comprehensively thought out, is exactly that - only a possible scenario. Of course then the challenge becomes how much to spend on your hedging strategy and how to apply it. Well, good risk managers are going to more than earn their money in the coming years, and the best of them will be the ones who can get the most bang for their company's buck in terms of hedging their business strategy. If you can get this right, it will have the potential to catapult your company's brand beyond your competition when the next wave of volatility hits the market as you will sail through it as opposed to being sunk.

Happy hunting,
Andy Shaw