Stock Price Jumps, Predictability Of Return And Volatility

This paper studies the jump behavior based on the A+H cross-listed stocks. With a sample including69stocks cross-listing in the Mainland market and the Hong Kong market for the period before November30,2011. we study three problems:the differences of jump phenomenon between A shares and H shares、how to predict return and realized volatility using jump.Firstly, through the data, we find that:there are many differences in the price and volatility of the stock price between A shares and H shares, common jump days account for only40%, it means that most of jumps are inconsistent between A shares and H shares. Since the jumps of single stock price are more related to the companys’ unpredictable information and the comovement of two markets, even if there is no common jump, the jumps of H shares may have impact on A shares, the jumps of A shares also may affect H shares.Secondly, we build a return forecasting model which including jump factor that in the form of dummy variable. We find that the model is perform excellent not only in the sample, it also can predict future return accurately out-of-sample. Comparing the estimation result, it shows as follows:Firstly, A shares are more easily affected by H shares; Secondly, no matter the jumps of local market or the jumps of another market, the relationship between the jump factor and return is significant, and the jumps of local market have a greater influence on return; Thirdly, for the same return forecasting model, it is fit better for A’share return, it suggests that investors in A-share market can get more useful information form H-share market, the price discovery of H-share market is stronger. In the end, we also build an investment strategy according to our return forecasting model and calculate the three-month return base on the strategy, we find that our return is far greater than the buy and hold return and average three-month return. So, in reality, our model can play a guiding role on investors when they buy A+H cross-listed stocks.Finally, in the model of realized volatility, we adding jump volatility of another market in HAR-RV-CJ volatility forecasting model, we find that its goodness of fit is better than original HAR-RV-CJ model. With the increase of forecasting length, our model performs more and more better. Comparing the estimation results of two markets, we also find that A share market are also more easily affected by the jump volatility of H share market, the future price volatility of mainland market will follow the price change in the past month of Hong Kong market. This is because that Hong Kong market is more mature and transparency of information is higher, so it is more persuasive for investors in mainland market.