Abstract

In this paper, we develop ARMA-GARCH type models for modelling volatility and financial market risk of shares on the Johannesburg Stock Exchange under the assumption of a skewed Student-t distribution. Daily data is used for the period 2002 to 2010. Several GARCH type models are used including threshold GARCH, GARCH-in mean and exponential GARCH. The results suggest that daily returns can be characterized by an ARMA (0, 1) process. This means that shocks to conditional mean dissipate after one period. Empirical results show that ARMA (0,1)-GARCH(1, 1) model achieves the most accurate volatility forecast. These results are useful to financial managers and modellers in both emerging and developed economies.

Key words: GARCH, volatility clustering, risk, forecasting.