Intraday and Overnight Causality in Time and Frequency Domains: Evidence From Stock Returns and Volatility
研究了标普500、DAX和SSE指数隔夜与日内收益（波动率）之间的非线性因果关系，发现双向因果关系，且交易量是重要传导渠道，对投资者短期和长期收益与风险管理有启示。
ABSTRACT This paper investigates the causal links between the overnight returns (volatility) and the intraday returns (volatility) series of the S&P 500, DAX and SSE by using a new multiscale nonlinear Granger causality analysis framework. We discover a bidirectional and nonlinear Granger causality between the two series, indicating that robust predictive information can be derived from the other series. To explore the driving force behind nonlinear causality, the original returns (volatility) series are analysed through multiscale decomposition with ensemble empirical mode decomposition (EEMD), fine‐to‐coarse reconstruction and nonlinear Granger causality testing. By counterfactual inference in the time‐frequency domain, we find that statistical causality is driven by specific time‐frequency component(s) of the original series. We also find trading volume serves as an important transmission channel for nonlinear Granger causality. These observations suggest both short‐ and long‐term investment implications, providing meaningful insights for equity market investors in terms of yield and risk management.