Gold and tail risks

In this study, we consider as a predictor of gold return predictability, an alternative measure of systematic risk using the tail risk obtained from the four variants (Adaptive, Symmetric absolute value, Asymmetric slope and Indirect GARCH) of the Conditional Autoregressive Value at Risk (CAViaR) of Engle and Manganelli (2004). We conduct distinct analyses for the gold-tail risk nexus for both 1% and 5% VaRs across the in-sample and out-of-sample forecasts. The results of the in-sample predictability indicate contrasting effects of own tail risk and oil tail risk (a proxy for global risk factor) with negative and positive effects, respectively on gold returns reinforcing the safe haven property of the gold market against global risk. Evidence of the out-of-sample predictability supports the inclusion of both own tail risk and oil tail risk over the benchmark model and single-predictor (own tail risk) model for improved out-of-sample forecasts of gold returns. The results leading to these conclusions are robust to alternative proxies for oil price and magnitudes of VaR.