Gold prices and equity market crises: how accurate are the forecasts from a nonlinear model?

Abstract

This paper examines the role of gold as a hedge during financial crises using daily data for the period 1976-2015. Although it is known that gold prices tend to increase during equity market crashes and that price volatility is symmetric or, at most, exhibits positive asymmetry; relatively little is known about the nonlinear nature of this behaviour. In fact, it seems that the magnitude of cross-correlations and the degree of shock’s persistency not only varies over time but also changes from the pre-crisis and the post-crisis period. This behaviour suggests that structural changes play an important role in this process of adjustment and hedging. Therefore, we use a combined STAR-IGARCH (Smooth Transition AutoRegressive – Integrated GARCH) type model to obtain forecasts of gold prices and their underlying volatility according to the nature of each crisis event. These results are compared with those obtained from the traditional IGARCH volatility specifications and the encompassing forecast accuracy is tested. The analysis is carried out over the entire sample period and over subsamples obtained from the succession of main crisis that occurred between 1976 and 2015. Our findings show that the STAR-IGARCH forecasts outperform the traditional IGARCH volatility forecasts in most cases and that the consideration of fractional persistency improves the quality of our forecasts. The role of gold prices as a hedge, even under the presence of structural changes, is thus confirmed by our results.