Abstract:
We analyze the predictive effect of monthly global, regional, and country-level financial
uncertainties on daily gold market volatility using univariate and multivariate GARCH-MIDAS
models, with the latter characterized by variable selection. Based on data over the period of July 1992
to May 2020, we highlight the role of the global financial uncertainty factor in accurately forecasting
gold price volatility relative to the benchmark GARCH-MIDAS-realized volatility model, with a
dominant role of European financial uncertainties, and 36 out of the 42 regional financial market
uncertainties. The forecasting performance of the global financial uncertainty factor is as good as
an index of global economic conditions, with results based on a combination of these two models
depicting evidence of complementary information. Moreover, the GARCH-MIDAS model with global
financial uncertainty cannot be outperformed by the multivariate version of the GARCH-MIDAS
framework, estimated using the adaptive LASSO, involving the top five developed and developing
countries each, chosen based on their ability to explain the movements of overall global financial
uncertainty. Our results imply that as financial uncertainties can improve the accuracy of the forecasts
of gold returns volatility, it would help investors to design optimal portfolios to counteract financial
risks. Also, as gold returns volatility reflects financial uncertainty, accurate forecasts of it would
provide information about the future path of economic activity, and assist policy authorities in
preventing possible economic slowdowns.
