Forecasting crude oil volatility with geopolitical risk: Do time-varying switching probabilities play a role?

This study examines whether geopolitical risk (GPR) exhibits an ability to forecast crude oil volatility from a time-varying transitional dynamics perspective. Unlike previous studies that assume an oversimplification of the fixed transition probabilities for crude oil volatility, we develop an asymmetric time-varying transition probability Markov regime switching (AS-TVTP-MS) GARCH model. In-sample estimated results show that GPR yields strong evidence of regime switching behavior on crude oil volatility and that the negative shocks of GPR result in greater effects on switching probabilities than positive shocks. Out-of-sample results indicate that the AS-TVTPMS GARCH model containing the GPR index outperforms other models, suggesting that the consideration of GPR information and time-varying regime switching together results in superior predictive performance. Moreover, the predictability of oil volatility is further verified to be economically significant in the framework of portfolio allocation. In addition, our results are robust to various settings.

Automated summary

This study uses an asymmetric time-varying transition probability Markov regime switching (AS-TVTP-MS) GARCH model to investigate the predictive ability of geopolitical risk on crude oil volatility. The findings suggest that negative shocks of geopolitical risk have a greater impact on switching probabilities than positive shocks, and the model containing the geopolitical risk index outperforms others in predictive performance, indicating that considering geopolitical risk information and time-varying regime switching together leads to superior predictions.