Empirical Study Based On The Realized Volatility Of Chinese Stock Market Heterogeneity

As an indispensable component of financial markets, China Stock Market plays the role of weatherglass of a country's economy. Studies on China stock market do great help to the smooth and healthy development of China financial markets. Thanks to the core role of the returns and volatility, this thesis is focused on the research of realized volatilities, finding the evidence of the heterogeneous market hypothesis through HAR-RV model. We hope that the research of this thesis would help to understand our security markets more deeply.Firstly, this thesis selects the best approach to computing the realized volatility in order to provide accurate data for the HAR-RV model building. The realized volatility is an ex post measure for the latent volatility. However, if the raw trading data are used for the computation of the realized volatility, biases would occur due to the microstructure effects when the sampling frequency is too high or due to the missing information when the sampling frequency is relatively low. Three approaches are applied to estimate the realized volatilities of individual stocks listed on Shenzhen Stock Exchange, which are first-order bias-correct, exponential moving average and method of looking for a fit frequency, respectively. It is shown after comparison that the biases of the realized volatility can be effectively corrected by first-order bias-correct with tick-by-tick data. Therefore, the real volatility can be estimated based on the highest frequency data base in Chinese stocks markets.Secondly, this thesis tests the performances of HAR-RV model. The realized volatility of stocks characterized with long-memory, this thesis, using the daily realized volatilities of 39 stocks, compares HAR-RV model with AR(1)-RV, AR(3)-RV, ARMA-RV, ARFIMA-RV and GARCH models, and finds that the HAR-RV model possesses the best fitness, out-of-sample predictability and the best robustness through three forecasting measures of RMSE, MAE and MAPE as well as three criteria presented in this thesis.Lastly, based on the HAR-RV model, this thesis tests the heterogeneity of China stock market. According to the heterogeneous market hypothesis, the most striking feature of financial market is that different agents have different time horizons and frequencies of trading. Thus, market could be simply composed of short-term, medium-term, and long-term traders. The empirical results indicate that there do exist at least these three types of traders in China Stock Market. Short-term and medium-term traders are predominant in most of the stocks while the long-term traders may seldom participate in the stocks with small market capitalization or in some particular industries. Moreover, the non-tradable shares, which is the exclusive characteristic of China stock market, does not affect investment strategies of traders of different horizons.