MODELLING AND FORECASTING VOLATILITY OF EAST ASIAN NEWLY INDUSTRIALIZED COUNTRIES AND JAPAN STOCK MARKETS WITH NON-LINEAR MODELS

This paper explores the forecasting performance of several non-linear models, namely GARCH, EGARCH, APARCH, which further utilize three distributions, namely the Gaussian normal, the Student-t and Generalized Error Distribution (GED). In order to evaluate the performance of the competing models we used the standard loss functions such as the Root Mean Squared Error, Mean Absolute Error, Mean Absolute Percentage Error and the Theil Inequality Coefficient. Our results show that the asymmetric GARCH family models are generally the best for forecasting daily stock returns in the NICs countries. We also find that both Root Mean Squared Error and Mean Absolute Error forecast statistic measures tend to choose models that were estimated assuming the normal distribution, while the other two remaining forecast measures privilege models with t-student and GED distribution.