Can CBOE gold and silver implied volatility help to forecast gold futures volatility in China? Evidence based on HAR and Ridge regression models

Abstract

The main purpose of this paper is to detect whether the CBOE gold and silver ETF (implied) volatility indices, i.e. GVZ and VXSLV, can help to forecast the realized volatility (RV) of gold futures price in China from both in-sample and out-of-sample perspectives. The empirical results based on various heterogeneous autoregressive (HAR) models and Ridge regression method show significant better predictive performance in those models incorporating CBOE GVZ and VXSLV indices than others without them. In addition, the model shrinkage method, Ridge regression, is found to be superior to other HAR-type models in forecasting China's gold futures volatility by reducing the problems of overfitting and multicollinearity in various volatility predictors.