| With the development of market microstructure theory and the extensive use of high frequency data, microstructure noise and its effect on asset price behavior has attracted more and more attention. It has been a research hotspot of market microstructure theory in recent years.Considering the specified characteristics of Chinese stock market, this paper studies the characteristics of microstructure noise and the effect on price behavior theoretically and empirically. The study is based on market microstructure theory and ultra-high frequency data. The analysis consists of five main parts:1. This research is about the measurement of microstructure noise and the characteristics of noise. When price data were sampled at very high frequency, the sample moment of observed return is a consistent estimator of the moment of noise with the same order. Therefore the secondary moments of noise, as the measurement of noise, were estimated using data sampled at the highest frequency. Using this measurement, the characteristics of noise were investigated in Chinese market. It is showed that, the order of magnitude of noise was 10-6, much smaller than volatility. The noise had a downward trend after 2006 and varied inversely with the market trend. Moreover, noise was inversely proportional to stock size and its sectional distribution was skewed to right. Also, noise in Chinese market was higher compared to US market. At last, the relationship of noise and jump was analyzed using VAR and it is found that noise and jump do not have certain relationship.2. Research on the structure of microstructure noise. Firstly, the mechanisms of some important factors in noise were analyzed theoretically. Then noise is decomposed into different components according to time-variability of noise and its relation with information. Lastly, the relations of noise with liquidity cost and information asymmetry cost were analyzed using MRR model. It is showed that the two components can interpret more than 80% of noise and the effect of the latter is a little larger than the former.3. Research on the relation between noise and volatility of efficient price. After analyzing the effect of noise on volatility estimation, microstructure noise and the real volatility were extracted from the ultra-high frequency data simultaneously using the method of optimal sampling frequency. Then, the relation of them was investigated. The result showed that coincident and lagged volatility have positive effect on noise, however the effect is decreased with the increase of lag order. At last, the risk pricing ability of market noise and market volatility were investigated. It is found that market noise is not systematic risk factor in Chinese market, but the unexpected change of market volatility is.4. Research on market closing and intraday price behavior. The market closing effect-a special component of noise was investigated through the study of intraday price behavior. Firstly, the noise, public information and private information components in the price were extracted using the method of Beveridge-Nelson and SVAR model. The result showed that there are more public information and private information and higher noise in the first half hour after market opening because of large amount of overnight information, which is contrast with the last half hour before market closing. Then using the method of WPC to compare the price discovery of call auction and continuous trading periods in A-share and Hongkong market. The result showed that the discontinuity of trading induced by the mechanism of market opening and closing imposed great influence on market closing effect and intraday price behavior. |
PDF Full Text Request