Erb and Harvey, 2006). Furthermore, these results seem to indicate that the link between
gold and silver, again investigated in several articles (eg. Ciner, 2001; Georglev, 2001), is
weak. Overall, these findings present evidence against the use of gold and silver as a
hedge against similar risks in stock portfolios. However, we do detect a feedback,
bivariate volatility spillover, relation between gold and silver markets that might be
exploitable by option traders, who are more concerned with volatility movements
between assets than simply changes in prices between assets.
(Insert Table 4 about here)
(c) Robustness Analysis
In this section, we conduct a subperiod analysis to determine whether the conclusions are
robust in different time periods. We divide the sample into two subperiods and report the
findings from the first half in Table 5 and the second half in Table 6. Importantly, the
findings are somewhat different relative to the full sample analysis, with important time-
varying properties evident in the impact of the macroeconomics variables on the precious
metals returns. For example, we detect that volatility in the gold market is impacted by
conditional volatility of the dollar index in the first subperiod (but not the second),
although the first subperiod finding is largely consistent with the view that gold is a
financial asset.
(Insert Table 5 and Table 6 about here)
We also find that palladium volatility is sensitive to volatility in financial markets, in
particular the conditional volatilities of S&P 500 returns and its dividend yield, which is
13