An Empirical Study On The Portfolio Decision Based On Second Order Stochastic Dominance

Following the continuous development of stock market in China and the rising of investment demand caused by the increasing of disposable personal income, the requirement of maintaining and appreciating the assets value increasingly improve. Meanwhile, the environment for financial investment changes, and people’s willingness to invest increase. It makes the efficiency of investment tools become one of the problem that investors pay the most attention to. Modern investment theory is based on quantitative theory. It build portfolios with the rigorous way of thinking and performs its efficiency by the indexes that refer to returns and risks. Along with the completing of financial market, these quantitative theory also develop at a amazing extent. And one of the important theories is stochastic dominance theory. It make considerable development aboard, and can successfully simulate the financial market. As a part of the portfolio theory, it’s worth to be researched.The purpose of this thesis is to find the value of second order stochastic dominance model in China. We can prove the efficiency of this model in investment activity by comparing second order stochastic dominance portfolios and bench, making horizontal comparison that can find the difference between second order stochastic dominance portfolios and portfolios under other risk preference, and conducting a empirical analysis with the data of financial market in China.First, the thesis make a brief introduction to the background and objective of the research on portfolio, the development of the investment theory and the history of the research on stochastic dominance portfolio. Then we make a warm-up for customers’ risk preference theory and discuss the theoretical basis of stochastic dominance theory. We also narrate the characteristics of its risk preference and the condition for portfolio optimization. Furthermore, we make a brief discussion on the other relevant theories.This thesis adopt two of the most typical second order stochastic dominance theories, stochastic dominance model based on efficient sets and stochastic dominance model based on utility function, which are often used by foreign researchers. It builds some optimal portfolios, and upgrades the classical mean-variance theory with the concept of second order stochastic dominance. And it establishes the correctional stochastic dominance, then researches into its practicability. At the same time, we build a classical mean-variance model and a experimental portfolio as frame of reference, and bring it into the evaluation system built in this thesis. The performance evaluation system include two dimensions, one of them is in sample and out of sample and the other one is indicative analysis and second order stochastic dominance analysis. And then we will make a overall estimation according to both dynamic change and static state.The main part of the thesis is an empirical analysis about portfolios above. We chose 18 secondary industry data and market data to archive optimization. Then we calculated the evaluation indexes from both in sample and out of sample, the two aspects, thereby confirming the dominant relationship. Finally, we found that the second order stochastic dominance portfolio can not only locate in the feasible region of portfolios but also be included in the efficient set of portfolios. And it perform equal to the other portfolios,sometime significantly good for the increasing of portfolios’ return. However, for mitigating risks, it can’t perform better than other portfolios. And comparing with stochastic dominant relationship, second order stochastic dominance portfolios have a better performance. This approach is more useful for an investor whose risk-aversion level is high.