An Empirical Research Of The Conditional CAPM Based On The Learning Process

Since 1980 s, there have been some financial anomalies rising in the stock market that the traditional CAPM can not explain, such as size premium and value premium, thus the theory of traditional CAPM is challenged. Despite the modifications and expansion carried out by the foreign scholars on the traditional CAPM, the empirical study on the application of the modified CAPM did not achieve an ideal result due to the fact that the financial anomalies in Chinese stock market are more serious which is caused by the late start. Meanwhile, when adapting the economic theories of the developed countries into Chinese market, some incongruities would appear. For instance, China’s stock market has a large number of state-owned enterprises which are very different from those private companies in western countries in terms of the ownership and operation. Therefore, even though the expanded CAPM may better analyze the empirical statistics in Chinese stock market, the economic principles can hardly be explained.This paper presents a conditional CAPM on the basis of a learning process in the hope of solving the financial anomalies that the traditional CAPM can not interpret, from which superior empirical results can be derived compared with those derived from other state-space models. This paper adopts the Kalman filter by incorporating the unobserved variables into the model to conduct empirical analysis with regard to the data collected from Chinese A-share market. According to the empirical results, the writer carefully draws a conclusion that the issues of size premium and value premium that the traditional single-factor CAPM can not solve can be settled by the model, in which the learning process in line with the Bayesian rule is integrated with the conditional CAPM, even though the single-factor form of CAPM is retained.