Long memory in conditional variance

Abstract

SUMMARY Presence of long memory in return and volatility of the spot price of gram in Delhi market has been investigated. GPH method resulted strong evidence of long range dependence in the volatility processes for the series. Accordingly, FIGARCH model has been applied for forecasting the volatility of gram price. GARCH model and several extensions of GARCH models such as TARCH, EGARCH, Component GARCH and Asymmetric component GARCH have been applied for modelling and forecasting of return series. Evaluation of forecasting has been carried out separately in six moving windows by the help of mean squares prediction error (MSPE), mean absolute prediction error (MAPE) and relative mean absolute prediction error (RMAPE). The residuals of the fitted models were used for diagnostic checking. Diebold Mariano test was conducted for different pairs of models to test for the difference in predictive accuracy. It is found that FIGARCH model has better predictive accuracy as compared to all other models. It is also observed that component GARCH and asymmetric component GARCH models have better predictive accuracy than GARCH, TARCH and EGARCH models whereas there is no significant difference in the predictive accuracy of GARCH, TARCH and EGARCH models. The R software package has been used for data analysis