Abstract
M.Com. (Finance)
Predicting share return volatility accurately in financial markets has become increasingly important, as it offers investors, market analysts and risk managers the ability to correctly price financial instruments, effectively manage portfolios and conduct accurate risk assessments.
The most popular method to predict this volatility is by using GARCH models, but there is no consensus as to which model offers the most accurate forecasts for volatility. Factors such as sample size, country characteristics and the leverage effect all have an influence in determining which model delivers the most accurate forecasts for volatility of share returns.
The goal of this study is to determine how accurately share returns can be predicted by using GARCH, GJR GARCH and EGARCH models as well as a benchmark ARMA model. In-sample and out-of-sample forecasts will be conducted to determine which model offers the most accurate forecast of share returns on the JSE Top 40 Index.
Results indicated that the ARMA (1,2) model produced the most accurate in-sample forecast, while the asymmetric EGARCH (2,1) produced the most accurate out-of-sample forecast. These findings are consistent with those from other South African studies.