Examining the predictive information of CBOE OVX on China’s oil futures volatility: Evidence from MS-MIDAS models

Abstract

This study evaluates whether CBOE crude oil volatility index (OVX) owns forecasting ability for China’s oil futures volatility using Markov-regime mixed data sampling (MS-MIDAS) models. In-sample empirical result shows that, OVX can significantly lead to high future short-term, middle-term and long-term volatilities with regard to Chinese oil futures market. Moreover, our proposed model, the Markov-regime MIDAS with including the OVX (MS-MIDAS-RV-OVX), significantly outperforms the MIDAS and other competing models. Unsurprising results further confirm that OVX indeed contain predictive information for oil realized volatility (especially significant and robust in middle-term and long-term horizons) and regime switching is useful to deal with the structural break within the energy market. We carry out economic value analysis and discuss OVX’s asymmetric effects concerning different trading hours and good (bad) OVX, and find OVX performs better in day-time trading hours and the good OVX is more predictive for the oil futures RV than the bad OVX. The further discussion also confirms our previous conclusions are robust during the highly volatile period of the COVID-19 pandemic.

Intra-day co-movements of crude oil futures: China and the international benchmarks

“…( 2020 ) indicated a sharp increase in the risk spillover between Chinese and international crude oil futures, and similar findings are also evidenced by Lu et al. ( 2020 ). Under a context of high-risk fluctuations and strict market regulation, below we evaluate the robustness of pairs trading strategy based on the intra-day co-movement patterns over day trading sessions from January 2 to April 30, 2020.…”

“…Moreover, the INE's regulators suspended the night trading session during the first outbreak of the pandemic in order to avoid a catastrophic spillover effect from the global crude oil market. Despite the regulation, Yang et al (2020) indicated a sharp increase in the risk spillover between Chinese and international crude oil futures, and similar findings are also evidenced by Lu et al (2020). Under a context of high-risk fluctuations and strict market regulation, below we evaluate the robustness of pairs trading strategy based on the intra-day co-movement patterns over day trading sessions from January 2 to April 30, 2020.…”

Forecasting stock market volatility with a large number of predictors: New evidence from the MS-MIDAS-LASSO model

“…Since studies have found that the Beta polynomial is more suitable for time lags compared to other functions (Ghysels & Qian, 2019 ; Ghysels et al, 2007 ; Zhang & Wang, 2019 ), we define in Eq. ( 2 ) as a Beta polynomial which is usually used in the realized volatility forecasting researches (Ghysels et al, 2006 ; Lu et al, 2020 ; Ma et al, 2018 ): where f function in Eq. ( 3 ) is a Beta function which used to make sure the weight to be positive.…”

“…In this paper, in Eqs. ( 2 ) and ( 3 ) is set to be 1 following Ma et al ( 2019 ) and Lu et al ( 2020 ). The model described above uses the lagged RV as the predictor and thus can be used as a benchmark model to evaluate the relative predictive performance of our extended models.…”

“…First, compared with the widely used HAR-RV model, the MIDAS-RV model can better reflect heterogeneity because the estimation process of the MIDAS-RV model primarily utilizes a specific function to determine the weights of the multi-order lagged RV series, while the HAR-RV model directly calculates the equally weighted average of the 1-week lagged and 1-month lagged RV series (Ghysels et al, 2006 ; Lu et al, 2020 ; Santos & Ziegelmann, 2014 ). Second, modelling asset volatility using the MIDAS-RV model and its extensions can better reflect the multi-order lagged effects of potential predictors on high-frequency volatility with fewer parameter estimates than the HAR-RV model, and is less prone to overfitting problems (Ghysels et al, 2006 ; Lu et al, 2020 ; Santos & Ziegelmann, 2014 ). Thirdly, studies have proved that the MIDAS-RV model can obtain more valid prediction information from predictors and has better prediction performance than the widely used HAR-RV model (Ma et al, 2019 ; Santos & Ziegelmann, 2014 ).…”