COMPARING THE PERFORMANCES OF GARCH-TYPE MODELS IN CAPTURING THE BROAD INDEX VOLATILITY IN DHAKA STOCK EXCHANGE

This study conducted the empirical investigation for the volatility of Broad index of Dhaka stock Exchange (DESX) in Bangladesh. Asymmetric Generalized Autoregressive Conditional Heteroscedastic (GARCH) model was used for DESX index. According to Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC), GJR-GARCH (1,1) is found to be the most applicable model to capture the asymmetric volatility. Their performances were also compared under statistical error measurement tools, e.g., root mean squared error (RMSE), mean absolute error (MAE), mean absolute percentage error (MAPE), theil inequality coefficient and bias proportion analyses.