We first report the estimation results for equations (5) and (4) in Table 1, which are used
to obtain the conditional volatility estimates of the variables in our data set. The results
for equation (5), reported in the upper panel of Table 1, suggest that there is significant
dependency in some of the variables in our sample, such as the dividend yield on both
S&P 500 and World-ex-US indexes, and various macroeconomic variables including
money supply, industrial production and inflation. This is supported by both statistically
significant F-tests for joint exclusion of dependent variables in the equation and also, by
relatively large R-squared values. This is noteworthy since equation (5) focuses on the
predictability of the growth rate of the series. However there is little evidence for
monthly seasonality in the variables, again evidenced by the F-tests reported in the table.
Moreover, Ljung-Box Q-tests for autocorrelation at 24 lags do not indicate any remaining
dependency in the residuals of equation (5) for any of the variables.
Estimation results for equation (4), which focus on unconditional standard deviations of
the variables, are reported in the lower panel of Table 1. It can be observed that there is
dependency in unconditional standard deviations of precious metal prices, consistent with
the notion that there is a general dependency in volatility present in financial markets. In
addition, dividend yields, money supply and inflation series also exhibit dependency,
while there is little evidence of seasonality. It also appears that autocorrelation in the
series are fully accounted for, evidenced by the insignificant values of the Ljung-Box Q-
tests on the residuals.
(Insert Table 2 about here)
10