Global economic policy uncertainty and gold futures market volatility: Evidence from Markov regime‐switching GARCH‐MIDAS models

Abstract

This paper explores the effects of global economic policy uncertainty (GEPU) on conditional volatility in the gold futures market using Markov regimeswitching GARCH-MIDAS models. The in-sample empirical results suggest that GEPU indeed contains predictive information for the gold futures market, and higher GEPU leads to higher volatility within the gold futures market.Moreover, the novel model, which adds Markov regime switching with timevarying transition probabilities and the GEPU index, achieves relatively better performance than those of the other competing models from a statistical point of view. Furthermore, we discuss the asymmetric effects of different changes in GEPU on the gold futures market and the models' performances with different horizons, and we find that our new model has better predictive performance under negative changes in GEPU than under positive changes in GEPU. Further discussion also confirms that our previous findings are robust during two special cases, the global financial crisis and European debt crisis, during which the market suffered from fierce fluctuations and was fraught with considerable uncertainty. K E Y W O R D SGARCH-MIDAS model, GEPU, gold futures market volatility, Markov regime switching, volatility forecasting | INTRODUCTIONDifferent from financial assets, gold, as a real asset, is always regarded as an effective buffer to slow down economic and political shocks (

The role of economic policy uncertainty in forecasting housing prices volatility in developed economies: evidence from a GARCH-MIDAS approach

“…We choose the GARCH-MIDAS technique to combine the quarterly data of house prices with monthly data of EPU. This technique has been applied in numerous studies to address the volatility linkages between the variables series at different frequencies (Zhou et al , 2020; Ma et al , 2021). The novelty of this technique is that it incorporates data from both low and high frequencies (Engle et al ., 2013).…”

Forecasting stock return volatility using a robust regression model

“…The next three potential variables are uncertainty variables proposed by Ludvigson et al (2015). The last three variables are oil volatility, geopolitical risk uncertainty, and economic policy uncertainty (Christou et al, 2018; Li et al, 2020; Ma et al, 2020; Mei et al, 2020; Wang et al, 2018). The Clark and West (2007) (CW) statistic is used to test the statistical significance of the out‐of‐sample R 2 values.…”

“…The second set of variables includes three uncertainty variables proposed by Ludvigson et al (2015) and Jurado et al (2015): (13) financial uncertainty (FU), (14) macro uncertainty (MU), and (15) real uncertainty (RU). The last three variables are (16) oil volatility (OIL RV), (17) GPR, and (18) EPU (Christou et al, 2018; Li et al, 2020; Ma et al, 2020; Mei et al, 2020; Wang et al, 2018).…”

“…The second set of variables includes three uncertainty variables proposed by Ludvigson et al (2015) and Jurado et al (2015): (13) financial uncertainty (FU), ( 14) macro uncertainty (MU), and (15) real uncertainty (RU). The last three variables are (16) oil volatility (OIL RV), ( 17) GPR, and (18) EPU (Christou et al, 2018;Li et al, 2020;Ma et al, 2020;Mei et al, 2020;Wang et al, 2018). denote the actual volatility, the volatility forecast based on the model of interest, and the volatility forecast based on the benchmark model for the (t + 1)th month, respectively.…”

“…To understand and better illustrate the effect of potential variables on the performance of the ARH model, we further consider the extensions of the AR and ARH models, which include potential variables. We use 12 macroeconomic variables (see, e.g., Christiansen et al, 2012; Rapach et al, 2010; Welch & Goyal, 2008; Zhu & Zhu, 2013), three uncertainty variables (see, e.g., Jurado et al, 2015; Ludvigson et al, 2015), oil volatility (Wang et al, 2018), geopolitical risk uncertainty (GPR) (Mei et al, 2020), and economic policy uncertainty (EPU) (see, e.g., Christou et al, 2018; Li et al, 2020; Ma et al, 2020) as potential variables to check robustness. With positive values obtained for each of the additional variables, our results are robust to the addition of these potential variables.…”