An Empirical Test Of The Predictability Of China's A Stock Market Rate Of Return

In recent years,with the rapid development of China's economy,China's financial market has become one of the most important emerging financial markets in the world.However,the stock market as an important part of Chinese financial market,due to the policy guidance characteristics of the market itself,as well as the market started late,imperfect system and unreasonable structural design problems left over by history,the market mechanism cannot play its proper role.According to the efficient market hypothesis,this means that China's financial market has not reached the efficient market,especially the stock market.Therefore,in the past two decades,scholars at home and abroad have been scrambling to study the predictability of China's stock market and compare it with foreign markets.Therefore,this paper also studies the topic of stock market predictability.But,different from the previous literature,this paper constructs a tool variable estimation test method to reflect the real situation of the Chinese market.It is of great theoretical and practical significance to reunderstand the predictability of China's A-share market.The reason why this article build tool variables of stock returns predictability test,because in financial empirical models,especially the stock prediction model,we will regress stock returns on financial variables to test the stock earnings predictability,and without fully considering the variable characteristics of real data,resulting in statistical results being errors or unreliable.The main factor causing the result inaccuracy is the uncertainty of the continuity of independent variables.Because,the persistence characteristics of independent variables mainly include three processes: stationary,near unit root and unit root.Before performing regression with sample data,if the independent variable cannot be determined as a specific one of the above three processes and return directly,then the result must be inaccurate.Moreover,on this basis,when there is a high correlation between residual error of prediction variable and residual error of return variable,that is,endogeneity exists,the deviation of statistical inference result will be greater.Because the accuracy of statistical test results depends entirely on the assumption that we know the true continuity and endogeneity of independent variables in advance.Therefore,starting from this basic problem,this paper firstly constructs the tool variable estimation test method,and obtains the expression of the prediction variable coefficient estimator and its wald-test;Then,in view of the variable data used in the article,this paper download data from Guotai'an,Wind,China economic network statistics database and the people's bank of China's official website,and respectively use ordinary least squares method and instrumental variable estimation method after the reform of non-tradable shares(September 2005-December 2016)in the Chinese A-share market between stock returns and financial variables regression,in order to verify the accuracy of the latter;Finally,this paper further analyzes whether the predictability of China's stock returns will change with the change of the market through sample segmentation and grouping.Study found that: first of all,the consumer price index year-on-year growth rates,short-term rates(weighted average interest rate on interbank seven-day)and long-term interest rates(the 10-year Treasury yield to maturity)for the forecast of stock returns has obvious effect;Secondly,variable portfolios screened by AIC criteria(Akaike,1974)showed the best consistent predictive efficacy in different markets.In contrast,the predictive power of dividend ratio(D/P)and short-term interest rate combination is weak.Finally,the predictive test of long-term level tells us that the predictive ability of variables or combinations of variables exists,but it does not increase or even decrease with the increase of long-term level.The innovation of this article mainly includes the following aspects: first,the instrumental variable estimation methods can help us to correctly reveal the real relationship of dependent and independent variables,and then provide an effective tool to controll the market for macroeconomic management and decision makers;Secondly,the empirical test of China's stock market can help us to have a better understanding of the predictability of global stock market returns other than the US market.Finally,through the segmentation and grouping of sample test,this paper also can provide the certain reference value for investors in the face of different investment targets and how to choose reasonable and reliable predictor,and then get stock excess returns.