Do precious metal spot prices influence each other? Evidence from a nonparametric causality-in-quantiles approach

Highlights

• A two-way causality exists among precious metal prices.

• The results show differences in causality in mean and variance.

• Daily and weekly precious metal prices show different causality patterns.

• There is evidence of weak causality among precious metal prices in crisis and boom periods, indicating diversification benefits.

Abstract

Using a quantile causality approach, we examine the causal relationship among the spot prices of precious metals (gold, silver, platinum and palladium) through mean and variance. This methodology also allows investigation of the causality among precious metals during recessions, booms and normal market states. Employing daily spot price data from April 2000 to July 2016 we found evidence of bi-directional causality in mean and variance among the prices of precious metals. Results indicate a strong causality for the middle quantiles (normal time periods). Robustness of results is also examined by employing weekly spot price data. Overall our results have significant implications for policy makers, portfolio managers and investors.

Introduction

The safe-haven properties of precious metals have offered substantial impetus to policy makers and scholars alike to examine their multi-faceted behavior, especially as an alternative investment instrument. Studies in the recent past have suggested the favorable role of precious metals, particularly gold, in hedging and portfolio diversification strategies (Baur and McDermott, 2010, Reboredo, 2013a). Nonetheless, some studies also argue against such diversification benefits (Lucey and Li, 2015, Reboredo, 2013b). However, those studies which argue in favor of the use of precious metals for investments overwhelm those arguing the unsuitability of precious metals as a diversification avenue. Most of the studies have argued in favor of the usage of precious metals for investments, such as the following: safe investment target, a hedging tool against risk and inflation, and highly liquid investment, among others. This hedging property of precious metals is evident from earlier literature, such as that by Jain and Biswal (2016), who argue that investments in precious metals, particularly gold, greatly increase during economic shocks. Baur and McDermott (2010) also report that the nominal prices of gold rose by 42 per cent on the eve of the financial crisis, i.e., July 2007. Other scholars also report gold to be uncorrelated with financial assets during periods of high volatility or financial crisis (Baur and Lucey, 2010, Baur and McDermott, 2010), therefore making it an ideal hedging instrument.

Despite substantial empirical evidence on multifarious aspects of precious metals, there are few studies that investigate interactive and transitive behavior among them. Hammoudeh et al. (2011) examine the dynamics of correlation and volatility in price returns of gold, silver, platinum and palladium and suggest implications for risk management. Sensoy (2013) reports a one-way volatility shift contagion effect from gold to other precious metals and from silver to platinum and palladium. Thus, in this context, the literature encounters some pertinent questions, such as: (a) Is it only gold that dictates the prices of other precious metals? Or (b) do other precious metals (silver, platinum or palladium) also lead gold prices? Or (c) do precious metals influence each other's prices? These questions are intriguing, and to the best of our knowledge, are yet to be answered. Therefore, the objective of this paper is to investigate the causality among precious metal prices by employing the quantile causality technique proposed by Balcilar et al. (2016a). We believe such analysis will help investors and policy makers originate better decisions regarding precious metal price movements.

Though earlier literature emphasizes the relationship among gold, silver, platinum and palladium, they cannot be considered as a single asset class (Pierdzioch et al., 2016). Batten et al. (2010) analyzed the spillovers among four precious metals and suggested a weak integration between gold, silver, platinum and palladium. Interestingly, Lucey and Li (2015) argued that silver, platinum and palladium can exhibit safe-haven properties during times when gold loses its safe-haven characteristic. Agyei-Ampomah et al. (2014) revealed the ability of gold to hedge against losses in sovereign bond issues in the case of countries with serious debt problems. Furthermore, they suggested the hedging ability of metals other than gold in sovereign bond market losses during periods of jitters in financial markets.

The existing literature provides useful information on precious metal dynamics; however, little is known about dependence and causality among precious metal prices. Therefore, in this study, we use the recent causality technique proposed by Balcilar et al. (2016a) to investigate the predictability of one precious metal price by the prices of other precious metals through mean and variance. We employ daily spot price data of gold, silver, platinum and palladium for the period of April 1, 2000 to July 25, 2016. To check the robustness of our results, we employ the same methodology on weekly spot prices of precious metals over the same time horizon. Our results show a bi-directional causality among precious metal prices in mean and variance. However, the causality among precious metals varies to some extent between daily and weekly prices.

Our contribution to the literature on precious metals is three-fold. First, the non-parametric quantile approach allows us to consider all the market conditions at the same time (low or high volatility or any other economic shocks). Therefore, the approach allows us to investigate the conditions under which one precious metal price responds to other precious metal prices. Second, we consider both first (mean) and second moments (variance) to analyze the causality between prices of different precious metals. Precious metals may not have causality in mean but could have predictive powers in variance (volatility). Predictive power in volatility could be more useful for better portfolio diversification strategies. Third, we use the application of a recent methodology by Balcilar et al. (2016a) to analyze the dynamics among precious metals.

The rest of the paper is organized as follows. In Section 2 we provide a concise review of literature. The stochastic properties of the data are mentioned in Section 3. The estimation methodology is discussed in Section 4. The empirical results are presented in Section 5. The result of the robustness test is presented in Section 6. Finally, we conclude in Section 7.