What is value at risk?

B. Venkatesh

THE Securities and Exchange Board of India (SEBI) wants to introduce a margin system based on value-at-risk (VaR) model. What is VaR ?

VaR is the maximum loss you are likely to incur on your portfolio during a certain period. Suppose your model throws up a one-day VaR of Rs 10,000 on your portfolio of Rs 5 lakh at a 95 per cent confidence level. This means your one-day loss will be not more than Rs 10,000 in 95 out of 100 trading days.

VaR is only an estimate. This means there is a likelihood or probability of you losing not more than the VaR on most days. In the above example, we are 95 per cent likely to lose not more than Rs 10,000. This also means we are 5 per cent likely to lose more than Rs 10,000 on certain days. VaR, thus, depends on the confidence level.

If you want to be more conservative in knowing your likely losses, you have to increase the confidence level to, say, 99 per cent. Suppose, the daily VaR at a 99 per cent confidence level is Rs 20,000 on the above portfolio, it means you are likely to lose not more than Rs 20,000 in 99 out of 100 trading days. Notice that VaR increases as you increase the confidence level.

You decide on the confidence interval depending on what you want to measure. Suppose you want to measure your portfolio risk, a 95 per cent confidence level will suffice. Banks, on the other hand, will want to use a 99 per cent confidence level even for their investment portfolio because they need to adhere to the capital adequacy norms.

There are various methods for computing VaR such as parametric and Delta-Normal models. The simplest of the models is the parametric VaR, which assumes that the daily portfolio returns are ``normally distributed''.

In short, VaR is an estimate of the loss you are likely to suffer on your investment during a certain period. The VaR model, introduced by JP Morgan way back in 1980s, thus helps investors and professional managers manage their portfolio risks.