Financial stress and returns predictability: Fresh evidence from China

Our paper examines the implication of the China financial stress index (CNFSI) constructed by Park and Mercado (2014) on asset pricing. First, the CNFSI has a significant negative relationship with subsequent stock market returns, the statistics of in-sample R2 and out-of-sample ROS2 estimated by the predictive regression model are 5.078% and 5.881%, respectively. Second, from the statistical evidence, the predictive effect of the CNFSI is significantly better than that of popular macroeconomic variables and other existing financial stress indices for both in- and out-of-sample periods. Third, according to the results of a long-horizon analysis, the CNFSI has better predictive power for returns during bull markets than during bear markets. Finally, these interesting results are verified by a set of robustness tests. •Financial stress index (CNFSI) has a good negative predictability for stock returns.•The return predictability of CNFSI has greater than the two existing financial stress.•CNFSI is better at predicting returns in bull markets than bear markets.•These interesting findings are confirmed through a series of robustness checks.