The Effect of the Nikkei and the S&P on the All-Ordinaries: A Comparison of Three Models

Abstract

This paper examines the influence of shocks in the Japanese Nikkei Index and in the US S&P Index on the Australian All-Ordinaries Index. We present results from the application of three models-an autoregressive linear model, a GARCH-M model and a non-linear neural network model. According to standard forecast statistics, a restricted feedforward neural network model, incorporating parallel processing of information specified by time-zones, out-performs the linear and GARCH-M models. However, according to the Diebold-Mariano test of forecast accuracy, the mean loss differential between the neural network and the linear model is not statistically different from zero, while that between the neural network and the GARCH-M is statistically different from zero. © 1998 John Wiley & Sons, Ltd.