| The Traditional Mean-Variance Capital Asset Pricing Model (CAPM) by Sharpe (1964) and Lintner (1965) fares poorly in explaining observed cross-sectional average stock returns. Specifically, the value-weighted market portfolio of risky assets seems highly mean-variance inefficient relative to portfolios formed on factors like market capitalization (Banz, 1981), book-to-market equity (BE/ME) ratio (Fama and French, 1992), and momentum (Jagadeesh and Titman, 1993). These empirical results may reflect the limitation of the MV criterion. The"anomalies"which CAPM model cannot explain were evidence of inefficient market in certain degree, or the problems of CAPM model. However, the significance and persistency of the market anomalies means that the market maybe efficient. It is well known that asset returns cannot be described by mean and variance alone. For example, the monthly returns of many stocks exhibit positive skenwss and excess kurtosis. So we've found Stochastic Dominance Theory to replace MV model. In this paper, we demonstrated that China's stock market is Prospect Stochastic Dominance efficient through empirical research, and thus explore China's investors'behavior on investment.Some economists have successfully used Stochastic Dominance Theory to analyze investors'behavior, but few domestic researchers did this research. Thus there are some innovations in research methods in this paper.We use Stochastic Dominance Theory to test the efficiency of market portfolios. Stochastic Dominance Theory is a nonparametric method through possible results and related objective probabilities to analyze risky assets selection problems with no need of parametric specification and distribution assumption. Traditional Stochastic Dominance Theory is based on Expected Utility Theory as foundations of economics. Later General Stochastic Dominance Theory was developed, such as Second Stochastic Dominance (SSD), Prospect Stochastic Dominance (PSD) based on Prospect theory, and Markowitz Stochastic Dominance (MSD) based on Markowitz utility function. Stochastic Dominance Theory can help explain three potential problems of extended model: increasing marginal utility, probability transformation, and specification error. In this paper, we used the three different SD criteria to test market efficiency relative to benchmark portfolios formed on market capitalization, book-to-market equity ratio and price momentum. SSD adds assumption of global risk aversion, that is, utility is everywhere concave. PSD assumes an S-shaped utility function that is convex for losses and concave for gains. MSD assumes a reverse S-shaped utility function that is concave for losses and convex for gains.Then we described linear programming tests to test SD efficiency if investors are locally risk seeking with the first-order conditions. In addition, we did statistical inference to introduce the possibility of wrongly accepting PSD or MSD efficiency for an inefficient portfolio.We did empirical research on twelve years'data of China's stock market. According to the method of Fama and French (1992), we sorted and divided all stocks by market capitalization, book-to-market equity ratio and price momentum separately once a month or once a year, and computed the excess returns of each group. The results came after the test procedure on these excess returns. Our empirical results show that our stock market portfolios are PSD efficient, and the utility function of investors'preference appear S-shape. The result also means Prospect Theory can better describe investors'preference than traditional Expected Utility Theory. Individuals are risk seeking for losses and risk aversion for gains with typical feature of"loss aversion". Specifically, when investors are in a state of loss, the marginal utility came from fortune appear a decline trend, at this time, the expected utility of holding the security is much than selling the security, that is risk seeking.Our research also had some shortage, but we gave some useful suggestions for future research. Firstly, generalize the current SD tests by including additional necessary restrictions for global optimality. Our analysis focused on the necessary first-order condition, so the market portfolio is a local optimum rather than the global optimum for portfolio selection; Secondly, relax our economic assumptions (e.g., by considering the multiperiod consumption-investment problem, imperfect competition, and market imperfections like transaction costs and taxes), and relax our statistical assumptions (e.g., by using econometric time series estimation techniques to estimate a conditional CDF).Our research on China's investors'preference has significant meaning for establishing a long-term, healthy, and stable stock market. According to the features of China's stock market, on one hand, the expansion of our security market is by power of government; On the other hand, from the view of account number, most are small and middle investors whose decision behaviors will affect the development of security market to a large extent. By the end of this paper, we give some suggestions: Firstly, government should reduce intervene of stock market, but focus on the development of regulations of market. Government should pay more attention to the long-term regulation development rather than short-term price intervene, and thus leave space for investors to make their investment decision. Secondly, develop institutional investors, and impose on manage the institutional investors. Government should optimize the structure of investors, and guide institutional investors play an important role of market stable by benign competition. In addition, long-term investment vehicle should come out for soon to expand the group of institution investors. |
