An Empirical Study on the Stability and Time Variation of Betas in Shenzhen Stock Market

This paper used CUSUMSQ statistics of recursive regression and Kalman filter analysis based on state-space models to identify the stability and time variation of betas in Shenzhen stock market respectively. The research results show that: (1) the betas of all industries are unstable and have time-variation; (2) in the three used models, the betas of ten out of thirteen industries are fitted best by mean-reverting model, and the betas of the other three industries are fitted best by random coefficient model, however, the random walk model is not suitable to fit the betas of any industry. The conclusions show the limitation of traditional CAPM and provide a good reference for the further studies of CAPM.