Volatility Linkages Among Gold Futures in Emerging Markets

ABSTRACT

We aim to detect the cross-border volatility linkages among gold futures in emerging markets, which still remain an untapped area. China, India, Japan, Taiwan, Turkey, and U.S. futures markets are included in the sample. The volatility linkage analyses confirm the existence of volatility transmission among the majority of the sample countries’ gold futures. This article carries vital inferences and implications for policy makers and investors. The policy making is particularly important for China, which is a relatively isolated market. From investors’ perspective, the results indicate that the risk diversification and cross-market hedging opportunities in the emerging gold futures markets are quite limited.

KEY WORDS:

• emerging markets
• gold futures
• multivariate GARCH
• volatility spillover

Notes

1. Currently, gold futures are traded in twelve emerging market derivatives exchanges: Argentina, Brazil, China, India, Indonesia, Korea, Nepal, Philippines, Romania, Russia, Taiwan, and Turkey.

2. According to the 2012 World Gold Council Report, the daily average trading volumes in Japanese and U.S. gold futures markets are USD 1.8 billion per day and USD 20.8 billion per day, respectively.

3. The price series are collected from Bloomberg’s Database.

4. In the original data set, the gold futures price data started in January 2008 for all countries except Taiwan, for which the start date was September 2, 2008. Thus, to achieve consistency in the sample for all countries, the latter date was chosen as the start date.

5. Thomson Reuters, GFMS, World Gold Council.

6. Thomson Reuters, GFMS, World Gold Council.

7. If the excess kurtosis (Kurtosis-3) is less than zero, then the distribution is assumed to be platykurtic and it has shorter tails compared to a uniform normal distribution.

8. A detailed discussion of these models is excluded for brevity.

9. In model selection, the model with the highest absolute value in terms of information criterion is chosen. In this case, since all the information criteria have negative values, the model with the lowest negative value (VECH-MGARCH) corresponds to the highest value in absolute terms and thus is chosen as the best-fit model.

10. Diagonal GARCH (p,q) specification is not used in this study since one of the most notable drawbacks of this specification is that it ignores cross-variable volatility interactions, the focal point of this particular study.

11. Consistent with the model selection, the best-fitting distribution was selected according to Akaike, Schwarz, and Hannan-Quinn information criteria.

12. The gold lease rate is the difference between the gold forward rate and LIBOR. A positive gold lease rate induces traders to lend gold in the spot market or to take a long position in gold futures. Gold lease rate is also named gold interest rate.