Don’t miss the latest developments in business and finance.

Value at Risk: Forget the past

Image
Pablo Triana
Last Updated : Feb 05 2013 | 8:02 AM IST

Value at Risk is not dead. That’s surprising. This measure of likely potential trading losses, one of the keys to risk control at banks, has been tried and found desperately wanting. And yet the promoters of VaR think that a few tweaks are all that is needed to get it working again.

The VaR reform agenda varies. Some want the calculations, which apply estimated likely losses on the basis of historical patterns, to use larger data samples. A richer picture could be a more reliable past reflection of future patterns.

Or maybe less data would let recent events be absorbed into the model more quickly. Or perhaps there should be different weightings of data for different periods. Maybe the data needs be updated weekly instead of monthly.

No, think again. All these notions are like suggesting a new kind of plastic for an airbag which didn’t inflate during an accident. Such technical refinements won’t address the problem.

The problem with VaR isn’t in the historical data used, but in the unreliability of historical data. To start, VaR figures aren’t set in stone. Change the period and the VaR number shifts dramatically. The Basel II regulations on bank capital call for a minimum one-year observation period, but there’s no magic in that, or in two years, or two months. It’s an unstable foundation for risk control.

Then there are those unrealistic assumptions of the probability of exceptional events. VaR junkies seem to forget what it says in prospectuses for investment funds: “past performance does not guarantee future performance”.

Everyone now knows that markets don’t behave in the manner that statisticians describe as normal. Financial markets are in fact non-normal places, frequented by Extreme Events, often known as Black Swans. The intrinsically unpredictable cannot be anticipated, even with a million years of past data.

As regulators cast around for new ways to monitor trading risk at banks, they should just forget about VaR. The banks may complain that without this technique, proprietary trading wouldn’t work. If so, the answer is to get out of proprietary trading, not to make up VaR fantasies.

Also Read

First Published: Mar 19 2009 | 12:52 AM IST

Next Story