Conditional threshold effects of stock market volatility on crude oil market volatility
用条件阈值自回归模型研究股市波动对原油市场波动的非线性影响，发现该模型在预测下行风险时优于传统阈值模型，对投资者和政策制定者控制风险有参考价值。
In this paper, we analyze conditional threshold effects of stock market volatility on crude oil market volatility. We use Conditional Threshold Autoregression (CoTAR), a novel extension of TAR from a constant threshold to a time-varying threshold. The conditional threshold is specified as an empirical quantile of recent realizations of a threshold variable. This specification is expected to match investors’ relative perception of financial risk. The target variable is monthly realized volatility (RV) measures of the West Texas Intermediate, and the threshold variable is monthly RV measures of the S&P 500 Index. Our rolling window out-of-sample analysis indicates that the predictive ability of CoTAR is at least on par with TAR for all cases considered, and significantly better than TAR for some cases. The superiority of CoTAR is pronounced when the target variable is a downside RV measure. This is a useful finding which helps market participants and policymakers better control downward risks. • We study conditional threshold effects of stock market volatility on crude oil market volatility. • We use Conditional Threshold Autoregression, a novel time-varying threshold model. • The CoTAR specification matches investors’ relative perception of financial risk. • The target variable is monthly realized volatility measures of the West Texas Intermediate. • The predictive power of CoTAR is significantly better than TAR for some cases.