Precious metals, oil and the exchange rate: contemporaneous spillovers

ABSTRACT

We investigate the contemporaneous spillovers among precious metals, crude oil and the US$ exchange rate. We contend that conventional reduced-form vector autoregressive (VAR) models based on lead/lag relations do not fully capture the interactions among these series as these models ignore the contemporaneous effects. Using a Structural VAR model, we identify these contemporaneous spillovers, which are shown to be strong and asymmetric. We further show that not taking into consideration the contemporaneous interactions among these assets leads to inaccurate findings and inevitably to inaccurate interpretations of the causal relations among them.

KEYWORDS:

• Structural VAR
• Precious metal commodities
• Oil prices
• Impulse response functions

JEL CLASSIFICATION:

• O13
• Q30
• Q43

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ In related studies, Wang, Feng and Zhang (2014), Wang and Feng (2015), and Wang, Zhang and Liu (2016) focus on the relation between energy consumption and economic activity, where Wang, Feng and Zhang (2014) focus on the efficiency of energy use, Wang and Feng (2015) focus on the role of energy in production efficiency and productivity growth, and Wang, Zhang and Liu (2016) focus on carbon intensity. While these studies, in a sense, focus on spillovers among quantities (i.e. directional effects of one on another), our article focuses on spillovers among futures prices.

² Rigobon (2003) points out that the identification through heteroscedasticity is a very robust technique, as it is not sensitive to misspecification of the actual conditional volatility process. All that the technique requires is non-proportional shifts in the volatility of the residuals. In fact, the volatility process could be modelled as a multivariate GARCH process (Rigobon and Sack 2003; and Badshah, Frijns, and Tourani-Rad 2013) or estimation could be done using a regime-switching model (Lanne and Lütkepohl 2010).

³ The U.S. dollar index is a weighted index of the value of the U.S. dollar relative to a basket of major currencies.

⁴ Note that in 2008, the CME Group acquired NYMEX and electronic trading occurs on the GLOBEX system, which is the electronic trading platform of the CME.

⁵ To mitigate the effect of outliers, we remove any observations that, in absolute terms, are greater than seven times the standard deviation of the series.

⁶ Gold’s low volatility is consistent with the fact that gold has a monetary attribute, and a good portion of its demand goes to hoarding and part of its supply also comes from recycling.

⁷ We perform Breusch–Pagan tests for heteroscedasticity in the residuals of the reduced-form VAR. We find the residuals reject the null hypothesis of homoscedasticity at the 1% level for all series. These results are available on request.

⁸ For the sake of brevity, we do not discuss the full results of the VAR(1). However, we present the results of this model with the impact of macroeconomic news announcements in Table B1 in Appendix B.

⁹ Cumulative impulse response functions converge to a constant within 50 steps ahead.

¹⁰ We have examined various other windows as well and our results remain relatively unchanged.