Researching The Problem Of Forecasting The Coefficient Of Beta In Shanghai Stock Market

In the Capital Asset Pricing Model, the coefficient of beta is usually used to measure and analyze the risk, whereas the coefficient of beta estimated through historic data cannot represent the real value of the Beta of the future. So, it is important to the application of CAPM to reasonably and exactly forecast the Beta. Formerly, most of researches on the Beta based on the static model, but in China's stock market, which has been founded not long and without well system in all kinds of aspect, static model does not suit analyzing the Beta. So, it is necessary to search a new kind of way to forecast the Beta, which suits with the characters of China's stock market.Firstly, the thesis summarizes the literatures about the coefficient of beta, introducing the common ways of forecasting the Beta and analyzing the impracticableness of some conclusions of foreign researches in China.Secondly, considering the actual situation of the China's stock market and that the Beta has been proven not to be constant, the thesis starts with the definition of the Beta, analyzing the time varying variance and covariance of yield by Multivariable GARCH model, and then estimates the time varying Beta. Further analyzing shows that the series of the time varying Beta is a stationary time series even though it does not a constant.Thirdly, the thesis analyzes the Beta by ARMA model of unique variable and VAR model of multivariable respectively. The results of ARMA model analyzing show that most of the Beta of these stocks can be fitted by an AR(1), which supporting the conclusion of"regressing toward mean". With multi-variable model, the theses analyzes the speculation influences on the Beta, which shows that both of the turnover rate and the price/earning ratio Granger cause the Beta and that the VAR model excels the ARMA model on forecasting the Beta significantly. Furthermore, the impulse response function indicates that both of the impulse responses to the turnover rate and to the price/earning ratio do not have the same characteristics, the impulse responses to the turnover rate are consistent whereas the impulse responses to price/earning ratio are erratic, which inflect from the profile that there are too much speculations and that the information of financial affairs is very fuzzy in nowadays China's stock market.