Value at Risk

How do you weigh up the risks when you make an investment? In the financial world, the measure used to judge the pros and cons of a financial decision is called the "value at risk" (VaR). Rather obviously, it is only as reliable as the factors you plug into it, but this is often forgotten.

VaR summarises the expected maximum loss from a financial position during a given time and to a given level of confidence. For example, a risk manager who says that his position has a daily VaR of $20 million to a 99 per cent confidence level means that he expects the losses from the position to be higher than $20 million on no more than 1 in 100 trading days. This is hugely valuable to banking because it allows a great variety of risks to be described in comparable terms.

However, as this measure has come to be applied to more complicated financial instruments it has started to break down. What's more, the methods by which the VaR number is calculated - and their relative merits - are often ignored.

A number can be created using a variety of techniques, from historical time series to complex "Monte Carlo" mathematical modelling. The choice of methodology, underlying statistical data, and the modeller's implementation skills are all extremely important in determining whether the resulting VaR numbers are useful or misleading.