Estimation of Value-at-Risk using the bootstrap resampling method (A case study of the Tehran stock exchange)



Exponential growth of financial markets has revealed the importance of the Value-at-Risk (VaR), a renowned measure of market risk, more than past. Using normal-GARCH model is one of the basic methods for estimating VaR. However, the distribution of financial asset returns has more fat tail compare to the normal distribution. Therefore, in this study we implement a bias-correction procedure based on the bootstrap method to remove the deficiencies of the normal-GARCH model with respect to the appropriate VaR forecasts. Our results show that the correction procedure has improved the ability of normal-GARCH model in forecasting VaR for Tehran Stock Exchange Price Index (TEPIX), at least in extreme probability levels, and also VaR forecasts obtained from the t-GARCH model. Historical Simulation (HS) and Filtered HS models have been also studied to further compare the results of the bias-correction procedure.