Determinants of the Optimal and Symmetricity of the Volatility Model for Some Selected Nigerian Stocks

Abstract

Business data always exhibits certain degree of asymmetry and some heteroscedasticity which affect the models and model choice. This asymmetry can be improved to improve the forecasts performance of the model. This work considered cases whereby the errors in the model are normally distributed or normally violate assumption, to improve the efficiency in parameter estimation and forecast as well. The results show that to determine whether volatility require the normal inverse Gaussian distribution, non-normality, and asymmetry which very important in modelling financial returns. Common model selection criteria, the Akaike Information Criterion (AIC) and the Swartz-Bayesian Information Criterion (BIC), were used to determine which recommended model best fits the price and return series to choose. The results show that the best marginal values of the residuals of the GARCH model when estimating Dancem, GTCO, Vitafoam, Nestle, and Fidson were normal inverse Gaussian distributions, respectively.

Keywords

References