Rare disaster risks and gold over 700 years : evidence from nonparametric quantile regressions

Abstract

Using annual data on real gold returns and measures of rare disaster risks over the period of 1280–2016, we show the existence of nonlinearity and regime changes in the relationship between the two variables of concern, over and above the existence of non-normality in the data. In light of these issues, we rely on a nonparametric quantile regression model to show that real gold returns can hedge against such risks, but only when the market is in its bullish-state, with it being negatively impacted in its bearish-phase. Understandably, our results have important implications for investors seeking refuge in the safe haven of gold during rare disaster events. In addition, our findings, would require theoreticians to develop new asset pricing models, which would incorporate the state-specific impact of rare disaster risks on gold.