Time-Varying Beta In Chinese Equity Market

Capital Asset Pricing Model (CAPM) has become a cornerstone of modern financial theory. Even though, in empirical researches, the beta of CAPM are usual regarded as a constant. In this case, beta can be estimated by regressing firm return on market return under ordinary least squares (OLS).In an early research of Fabozzi and Francis (1978). it is indicated that beta are randomly time-varying. Various literatures tried to predict time-varying beta instead of the constant assumption. The major recent researches focus on exploring the determinant of time-varying Beta. The domestic researches are leading by this trend, though their attention is more on interpreting the descriptive statistics of time varying Beta. Therefore it is meaningful to study on the time-varying Beta of Chinese equity market by applying the developed theory of time-varying Beta. Exploring the secret of time-varying beta, not only helps the portfolio managers to better manage their market risk exposure, but also provide valuable insights for the policy maker.This thesis explores the determinants of time-varying beta for 33 Chinese sectors during 2000 and 2014. Bivariate BEK.K Multivariate GARCH model is used as measurement technique of time-varying beta. Based the analysis on the series of time-varying Beta, the author performs structural break test, and then researches the relations between time-varying Beta and market characteristics, volatility, and climate index. Additionally the author evaluates the sector dispersion under various circumstances. In the end. the author tests the fitness of GARCH model to time-varying Beta.The research arrives at the conclusion that during the sample period 2000-2014. most of the series of time-varying Beta presented structural break points, which possibly related to the reform and material changes in the general market condition. Sector Beta fluctuate differently with the market. When the volatility increases, it is likely the dispersion of sector Beta will be decreasing.Though the author arrives several useful conclusions in the above, there are mainly three inadequacies lying in the thesis. Firstly, the author merely has considered market beta. However market beta has very limited explanatory power on return. For further researches, more and different types of betas should be added in BEKK Multivariate GARCH model. Secondly, in the research on time-varying beta and market condition, the author classified bearish and bullish market according to market index and subjective adjustment. It is likely that different classification would generate different results. Lastly, because of lack of consensus theoretical basis in the field of time-varying beta, the conduction mechanism of bearish or bullish market is still left unclear. Thus the causality between market and time-varying beta is hard to explain.