Estimation of garch model using error distributions for accuracy measure of the Nigerian economy

Abstract

The study examined the performance of some symmetric and asymmetric GARCHmodels onNigeria’s economy using the real gross domestic product cost at a constant factor (RGDPCF).The series comprises all the monthly values of the data covering the period from January 2000 to December 2019. It filled the lacuna by explicitly modeling and comparing four different GARCH models for the variable mentioned above. The order of the GARCH models was specifiedand the comparison among all the models was done on the bases of the error distribution and model selection criteria. The result of the symmetric model GARCH in normal error distribution assumption was found to have the highest volatility of 150.1 percent persistence impact, however, GARCH in generalized error distribution assumption with the Akaike model Information Criterion(AIC=4.289) is considered the most appropriate symmetric model. Similarly, the TGARCH in Generalized error distribution assumption has the highest impact volatility persistence of 217.6 percent. The leverage effect of the TGARCH-Normal is -0.114 (p-value=0.000), andthat of TGARCH-GED is -1.247 (p-value=0.000,). Hence, the overall best fit model for the Real Gross Domestic Product at Constant Factor Cost (RGDPCF) based on the Akaike information model selection criterion (AIC = 4.065) is the TGARCH model in Generalized Error Distribution Assumption.