Does oil predict gold? A nonparametric causality-in-quantiles approach

Abstract

This paper examines the predictive power of oil price for gold price using the novel nonparametric causality-in-quantiles testing approach. The study uses weekly data over the April 1983-August 2016 period for both the spot and 1-month to 12-month futures markets. The new approach, the causality-in-quantile, allows one to test for causality-in-mean and causality-in-variance when there may be no causality in the first moment but higher order interdependencies may exist. The tests are preferred over the linear Granger causality test that might be subject to misleading results due to misspecification. Contrary to no predictability results obtained under misspecified linear structure, the nonparametric causality-in-quantiles test shows that oil price has a weak predictive power for the gold price. Moreover, the causality-in-variance tests obtain strong support for the predictive capacity of oil for gold market volatility. The results underline the importance of accounting for nonlinearity in the analysis of causality from oil to gold.