Gold and crude oil: A time-varying causality across various market conditions

Abstract

This paper examines the relationship between the returns and volatility of gold and crude oil markets using a novel procedure capturing both the quantile and time-varying Granger causality in mean and in variance, which is necessary to understand the complexity and multi-layered causal flow between these two strategic commodities. Using daily data from March 2nd, 2005, to February 7th, 2022, covering the global financial crisis and COVID-19 pandemic, the results are summarized as follows. Firstly, there is a bidirectional causal relationship in mean between gold and oil returns, regardless of the market conditions. Secondly, causality in variance varies across quantiles and mainly flows from gold to Brent during moderate to high volatility states of the volatility conditional distribution. Thirdly, the rolling-window causality-in-quantiles approach provides solid evidence that the return and volatility predictability between gold and oil markets is both time and quantile dependent, with evidence that the causal flows from Brent to gold are relatively stronger and span longer episodes than those running from gold to crude oil. The overall findings imply the unsuitability of applying a static quantile causality approach between these two strategic commodities, particularly over a long sample period, since the interactions between gold and crude oil markets tend to fracture suddenly under crisis periods. Accordingly, our main findings have important implications for portfolio diversification and hedging decisions.