Anmerkungen:
Nach Informationen von SSRN wurde die ursprüngliche Fassung des Dokuments January 5, 2012 erstellt
Beschreibung:
In the context of a continually changing and reforming financial market, stock market volatility plays a vital role in indicating macroeconomic environment changes, market participants' expectation and interaction mechanism. Market volatility research has been conducted by worldwide academics from different perspectives. However, different backgrounds, assumptions, methodologies and intentions have led to heterogeneous volatility models, making volatility forecasting a controversial topic. This paper tests the hypothesis that stock return distribution assumptions underlying Australian Securities Exchange's Volatility influence the performance of volatility forecasting. The methodologies mainly involve empirical analysis, for instance, three GARCH models in terms of GARCH-Normal (GARCH-N), GARC-Student-t (GARCH-ST) and GARCH-Skewed Generalised Error Distribution (GARCH-SGED) are applied. Not only the daily returns, realised weekly and monthly volatilities of S&P/ASX 200 Index and ASX All Ordinaries Index are calculated over the period of 10 years, but also the relative out-of sample volatilities among the three models are compared. The output indicates that based on the model selection criteria of MSE and MAE, the GARCH-ST is superior to either GARCH-N or GARCVH-ST over short-run forecast horizon while GARCH-SGED performs better than either GARCH-N or GARCH-ST over long-run forecast horizon. Next the DM-tests confirm these conclusions and demonstrate that modelling market volatility should take account of the negative skewness and leptokurtosis embedded in the stock return distributions. Then dynamic volatility dependence and volatility-correlation co-movement are identified and examined to solidify and extend current volatility research. Finally, the following risk management techniques and policies are suggested: high-dimensional volatility modelling and out-of-sample forecasting should be based on appropriate assumptions of stock return distribution to increase forecast accuracy; volatility dependence and volatility-correlation co-movement may reduce the benefits of stock diversification