The importance of global economic policy uncertainty in predicting gold futures market volatility: A GARCH‐MIDAS approach

• Author: Yichuo Qian,Libing Fang,Honghai Yu,Baizhu Chen
• DOI: http://doi.org/10.1002/fut.21897
• Published date: 01 March 2018
• Date: 01 March 2018

Received: 26 April 2017

|

Revised: 19 November 2017

|

Accepted: 19 November 2017

DOI: 10.1002/fut.21897

RESEARCH ARTICLE

The importance of global economic policy uncertainty in

predicting gold futures market volatility: A GARCH-MIDAS

approach

Libing Fang

1

|

Baizhu Chen

2

|

Honghai Yu

1

|

Yichuo Qian

1

1

School of Management and Engineering,

Nanjing University, Nanjing, Jiangsu, China

2

Marshall School of Business, University

of Southern California, Los Angeles,

California

Correspondence

Assoc. Prof. Honghai Yu, School of

Management and Engineering, Nanjing

University, #22 Hankou Road, Gulou

District, Nanjing 210093, China.

Email: hhyu@nju.edu.cn

Funding information

National Natural Science Foundation of

China, Grant numbers: 71301019,

71401071, 71472085, 71672081,

71720107001, 71771117

This paper applies the GARCH-MIDAS model to examine whether information

contained in global economic policy uncertainty (GEPU) can help to predict short-

and long-term components of the gold futures return variance. Our results show that

GEPU positively and significantly forecasts the future monthly volatilities for the

aggregate global gold futures market. The forecasting power of GEPU remains strong

in an out-of-sample setting. Moreover, further out-of-sample tests show that the

GARCH-MIDAS model with GEPU and realized volatility outperforms all other

specifications, indicating that including low-frequency GEPU information in the

GARCH-MIDAS model significantly enhances the forecasting ability of the model.

KEYWORDS

GARCH-MIDAS, global economic policy uncertainty, gold futures market, volatility forecasting

JEL CLASSIFICATION

C53, C58, G17

1

|

INTRODUCTION

Volatility modeling and forecasting is a crucial issue in risk management, asset allocation, pricing of derivative

instruments, and policymaking. Furthermore, volatility forecasting is an important task in financial markets which has

held the attention of academics and practitioners for many years. As the seminal papers of Engle (1982) and Bollerslev

(1986), a large body of literature has investigated the time variation of volatility and the factors behind it using different

econometric models (Cai, Chen, Hong, & Jiang, 2017; Chen, Jiang, Li, & Xu, 2016; Engle, Ghysels, & Sohn, 2013). In

particular, Ghysels, Santa-Clara, and Valkanov (2006) introduced mixed data sampling (MIDAS), which allowed

inclusion of data from different frequencies in the same model. As mentioned by Ghysels et al. (2006), the MIDAS

specification has several advantages. First, it is very flexible and captures a rich set of dynamics that would be difficult to

obtain using standard same-frequency regressions. Second, it is easy to extend it to nonlinear and multivariate settings.

Finally, the model can capture a component of the variance that is not priced by the market and consequently that is

unrelated to expected returns.

Based on this, Engle et al. (2013) proposed the GARCH-MIDAS model. It decomposes the volatility into the short- and long-

run component, one pertaining to short-term fluctuations and the other to a long-run component. The approach has the

advantages that it separates short- and long-run components of volatility and uses a direct approach imputing macroeconomic

time series to capture the economic sources of stock market volatility. More importantly, Engle et al. (2013) find that models with

the long-term component driven by macroeconomic variables are at par in terms of out-of-sample prediction and outperform

pure time series statistical models, in which the loss of efficiency due to multiple steps estimation is reduced. With this method,

J Futures Markets. 2018;38:413–422. wileyonlinelibrary.com/journal/fut © 2017 Wiley Periodicals, Inc.

|

413