A Test of Alternative Value-At-Risk Models During Volatile Periods

Abstract

This paper compared the performance of alternative models for estimating Value at Risk (VaR) of four different currencies against the Indian rupee. I examined whether incorporating a volatility estimate capturing the ARCH effects in the normal linear VaR model yielded a better estimate of market risk than the traditional models based on historical simulation and historical moving average volatility. I tested the effectiveness of different VaR models during the volatile period of June-September 2013 and found that VaR models based on an estimate of time-varying volatility performed better than traditional models during turbulent times.