Relative Signal Jump Variance, Investor Attention, and the Cross-Section of Chinese Stock Returns

We estimate the relative signal jump variance (RSJV) as the difference between the realized positive half-variance and negative half-variance divided by the realized variance using high-frequency intraday data and investigate its role in the cross-sectional pricing in the Chinese stock market. We find that the RSJV has a negative pricing effect. The stocks in the lowest RSJV quantile generate more than 1.30% risk-adjusted return per month compared to the highest RSJV quantile. This anomaly cannot be captured by other established asset pricing effects and remains robust to many considerations. Furthermore, we rule out the risk-based explanation and find this effect is partially driven by limits-to-arbitrage and limited investor attention