Risk management and statistical models

Hi,

http://business.timesonline.co.uk/tol/business/columnists/article5871007.ece

More articles appearing in the press about risk management and the failings of financial risk models. I've addressed some of these topics in a previous blog already but thought I'd comment on this article specifically. The authors are barking up the right tree but are missing a few key issues.

Risk models don't make mistakes, risk managers make mistakes. This is a little like the maxim, "guns don't kill people, people kill people". You could have all the possible statistical models in the known universe but they will be of little help as you'll probably just confuse yourself in a fog of model output and make even more mistakes. The more models you have, especially if they are very technical, the more bogged down in maintaining the model's integrity you will become and the less time you will have for managing your risk. A sweet spot for the number of risk models you need is a maximum of 2, minimum of 1. Pick a couple of models you understand well rather than commissioning the latest unproven research from a crackpot, neural networking, aerodynamicist turned risk manager. If you don't understand how a model works, how can you undestand when to use its output or when to ignore it?

Once you have your tools, then you're ready to play. And this leads onto the next key point. The clue here is in the name, 'risk manager'. Risk managers aren't risk observers. They are risk 'managers'. There are a variety of instruments out there for them to manage risks with. Risk managers earn their money by deploying clever and cost effective hedging strategies which support the company's business strategy.

Remember, markets are the collective will of all those people able to deploy capital on any particular day. Or, in alternative terms, the market will move in response to the volume of buyers and sellers in any particular product. No model in the world will ever capture this effect accurately given the idiosyncratic whims and flaws of human nature.

So once you are efficiently hedged, it doesn't matter how irrationally crazy the rest of the investing world gets, and it can get pretty crazy when you look at some of the recent peaks and troughs of certain assets. Through clever hedging you have turned your downside from an unexpected unkown, based on the whim of the market, to a number you know regardless of whatever scenario the market throws up. And businesses love known knowns. Just ask Rummy Rumsfeld.

Simple huh. Yet I haven't seen one article out there that outlines this point.

Happy hunting
Andy Shaw