August 2009

BACK OFFICE: Don't blame it on VaR

VaR has a misleading description. ‘Value at risk’ sounds like it is communicating the maximum rainfall rather than just an idea of whether a rainstorm is likely. Peter Ainsworth comments...

peter_ainsworth.jpgMichael Fish, the longest serving weather presenter on British television, became a subject of infamy in the wake of the Great Storm of 1987. A few hours before the storm broke in October 1987, he said during a forecast: “Earlier on today, apparently, a woman rang the BBC and said she heard there was a hurricane on the way... well, if you’re watching, don’t worry, there isn’t!” That evening, the worst storm to hit South East England since 1703 caused record damages and killed 19 people.

In the Great Credit Crunch of 2007-2009, trillions of dollars were lost due to excessive lending by banks which thought the economic weather would be much fairer than turned out to be the case. On this occasion the “woman who rang the BBC” is Nicholas Taleb, once a prop trader and subsequently author of The Black Swan, who wrote that “the government-sponsored institution Fannie Mae, when I look at its risks, seems to be sitting on a barrel of dynamite” among other prescient forecasts of the financial crisis.

Michael Fish got off lightly, but Taleb is making public his anger that his forecasts were not taken into account and he wants to punish the financial weather experts.

“My outrage,” he says, “is aimed at the scientist-charlatan putting society at risk using statistical methods.” He goes on to say that financial engineering “idiots” are dangerous to society and that a “crime” has been committed. “We would like society to lock up quantitative risk managers before they cause more damage.'' The gist of his message in an interview in Derivatives Strategy was that value-at-risk (VaR) is utterly useless as a risk management tool, elsewhere advising that the authorities should “remove [VaR] books from the shelves – quickly”.

So, why did Michael Fish get off so lightly and what is this VaR thing that is under attack?

With a weather forecast everybody, especially the British, knows it’s not meant to be accurate. Indeed, we joke about the inconsistency between weather forecasts and the outcome. Michael Fish, being a bit more inaccurate than usual, was simply funny, confirming what we already knew to be true – that weather forecasts are useful but you’d be an idiot to place total reliance on one.

VaR is simply a financial weather forecast. A high VaR suggests stormy weather and the risk of big losses, while a low VaR indicates a balmy day and rain, in the form of big losses, is not likely. But VaR, using its full name, has a misleading description. ‘Value at risk’ sounds like it is communicating the maximum rainfall rather than just an idea of whether a rainstorm is likely. Indeed, in a recent speech, the FSA’s Lord Turner implied that even he had been mislead when he said: “We know that [VaR] praised as a mathematically precise measure of risk.” But no professional statistician would describe VaR that way.

VaR is usually expressed in currency amounts. For example, a portfolio could have a VaR of £100m (€115.9m). In full that reads as ‘our Value at Risk is £100m’. It makes it sound like that’s all that could be lost. If, instead, and more appropriately, the name VaR was replaced by ‘best guess of how much we might lose on 99 days out of 100, but not taking account of the 100th day when we could lose a lot more’, it’s possible there’d be a better understanding of what VaR is communicating and less anger when losses do exceed the VaR measure.

In reaction to the controversy over the failure of weather forecasters in October 1987, the term ‘Michael Fish effect’ was coined, whereby British weathermen are now inclined to predict a worst-case scenario in order to avoid being caught out.  Current regulatory proposals for the financial sector seem very much in the same vein. While always anticipating the worst will reduce the risk that forecasts are out of line with reality, the cost of any such restrictions on investors and bank activity will likely be many years of lower-than-otherwise economic growth.

It would surely be a better solution simply to educate the investing and financial community about risk, whether measured by VaR or any other approach. Simply put, we need risk measures, but we have to treat them with the degree of uncertainty we naturally attach to a weather forecast. Don’t blame Michael Fish, or VaR!

• Peter Ainsworth is a managing director at EM Applications

©2009 funds europe