Question

You have been using a VaR model based on the normal distribution, with the variance estimated from the last 10 years of daily returns. You have found that it is not reflecting changing market conditions adequately. Outline two methods you could use you enhance your VaR model to be more adaptable through time, discussing the impact of any model features you may have to select.

Answer 1

1. Alter the distribution of returns to be non-normal

It is possible the the returns from the chosen asset might not be normally distributed and hence other non-normal distributions such as lognormal or gamma distribution should be considered which reflects the changed market conditions.

2. Consider other time period to model variance or add loadings to the variance.

Instead of last 10 years of daily returns we can use past returns data for other similar assets and compare the volatility of the returns. If might be also be possible to add loadings to the estimated variance to allow for the changing market conditions.