Modelling and forecasting the predictability of stock market return in asian countries by using hybrid arima-garch models

Abstract

Predictability of the stock market return has been a crucial topic over a decade. The ability to forecast and predict the stock market price allows investors to make investment decisions at the lowest risk and also allows policy makers to evaluate development of stock markets as to design rules and regulations. Thus, this study was conducted to serve two main purposes. First of all, hybrid models was developed between Autoregressive Integrated Moving Average (ARIMA) model and Generalized Autoregressive Conditional Heteroscedasticity (GARCH) family model for daily stock market data. In GARCH family models, there are GARCH, EGARCH and TGARCH where GARCH is symmetric model and EGARCH and TGARCH are asymmetric models. As hybridization of ARIMA model with different GARCH family models have different level of performances, each of the established hybrid models are evaluated using AIC, MAE, RMSE as well as MAPE to identify the outperformed model. In this study, daily stock prices of nine Asian countries (China, Hong Kong, India, Indonesia, Korea, Malaysia, Philippines, Singapore, and Thailand) are being used. EViews and R studio software act as the tools to perform the analysis. Results show that hybrid ARIMA-EGARCH model outperformed. On the other hand, identification of the calendar effects of all the nine Asian countries is the second concern of this study. The results show that each of the Asian countries have different calendar effects.