Research On The Impact Of Investor Sentiment On Stock Market Return Volatility Under The Binary Market Structure

The stock market is the core component of the capital market and it provides an important support for the development of the real economy by providing direct financing channels for enterprises in the real economy.However,The irrational investor sentiment has become a normal feature in China's stock market,and it will lead to greater uncertainty which will have a great negative impact on the healthy and stable development of China's stock market.Besides,the binary market structure containing mean and extreme states has become a normal feature of the operation of the Chinese stock market.Therefore,it is of theoretical and practical significance to explore the impact of investor sentiment on the volatility of China's stock market under he binary market structure.This dissertation takes the investor sentiment as the research topic.We designed an optimization method for investor sentiment proxy index,compared and analyzed the advantages and disadvantages of three methods to construct investor sentiment comprehensive index so as to improve the reasonable measure of investor sentiment index,which is the key problem to the study of investor sentiment.After that,it constructs the mathematical model to explore the influence mechanism of investor sentiment on the stock market returns to provide basis hypothesis for empirical research.On this basis,we empirically test the impact of investor sentiment on stock market returns under the binary market structure.The main contents and conclusions of this dissertation are as follows:(1)Deducing the influence mechanism of investor sentiment on stock market returns based on the improved noise trading theory model.Based on the DSSW noise trading model,this dissertation modifies the investor structure of the stock markets,increases the short selling limit,considers the systematic risk caused by the change of the fundamentals,and constructs a noise trading model which is more suitable for Chinese stock market.And based on the improved noise trading model,we deduced the mechanism of influence of investorsentiment on stock returns.Based on the modified DSSW model,we can deduce that the effects of investor sentiment on stock market return and its volatility are embodied in four effects: hold more effects,price pressure effect,Freedman effect and create space effect.Therefore,four hypotheses are proposed in this dissertation.(2)Selecting the proxy index data of the weekly frequency to construct an investor sentiment monitoring index system.In order to track the changes of investor sentiment on the stock market more accurately,this dissertation selects 20 sentiment proxies and 380 weeks' circular frequency data whose information granularity is smaller and has higher frequency,and then classifies them by different categories to construct investor sentiment index monitoring system.Aiming at the problem that the selection of proxy index of investor sentiment often has a strong subjective component,lacks verified instructions to rationality of the selected indexes and doesn't set a culling mechanism for unreasonable proxy indicators,this dissertation proposes a set of methods to optimize investor sentiment proxy variables.By using the optimization method,the sentiment proxy variables suitable for the construction of a comprehensive measure of investor sentiment index can finally be determined.(3)Construction of investor sentiment comprehensive index.Most of the current literatures use the PCA method to construct the investor sentiment comprehensive index.Considering the limitation of the principal analysis method,this dissertation adopts the PCA method,PLS method,and the LASSO regression method to construct investor sentiment index,and compares investor sentiment index constructed by three methods from three angles including rationality test,robustness comparison and the ability to predict the closing price of the CSI.The empirical results show that the LASSO regression method is more suitable for constructing investor sentiment comprehensive index.Its main advantage lies in that it can do variable selection and complexity adjustment while it fit the generalized linear model,so that when reducing the complexity of the model,we can minimize the impact of outliers.(4)Using OLS regression method and Quantile regression method to test the effect of investor sentiment on stock market returns under the binary market structure.This dissertation empirically tests the influence of investor sentiment on the domestic stock market under the homogeneous market environment and extreme market environment.By comparing and analyzing the impact of investor sentimenton the stock market,we can see whether the impact of investor sentiment on the stock market only exists in emerging markets.The results show that: firstly,in the mean market environment,there are both "hold more effect" and "price pressure effect" in the stock market of China and the United states.There are obvious"freedman effect" and "creation space effect" in Chinese stock market,while it is not obvious in American stock market.Next,in the extreme market environment,China stock market still exists obvious "hold more effect" and "price pressure effect",and"hold more effect" than the "price pressure effect",while the United States stock market in the "hold more effect" and "price pressure effect" is not obvious.Besides,only when the stock market returns are very low,the "freedman effect" will exist in China's stock market,when the stock market volatility Chinese is gentle,only the pessimistic investor sentiment volatility has a significant impact on stock market volatility;and there will never be “freedman effect" and "creative space effect" in the U.S stock market.Comparing the empirical results between two countries,the influence of Chinese investor sentiment on stock market returns and volatility is greater than that of American investor sentiment.The innovation of this article is reflected in the following three aspects.(1)Constructing monitoring system of investor sentiment index and proposing the optimization method of selecting investor index,this improves the selection of agent variables subjectively and without culling mechanisms.From the 6 market dimensions,the corresponding 20 week frequency sentiment agent indexes are selected,and the investor sentiment index monitoring system is constructed.For the first time,we select the best index of New Wealth as subjective sentiment indicators.This dissertation proposes an optimization procedure to screen investor sentiment proxy variables,which improved the situation that people are subjective when choosing sentiment proxy variables.(2)Applying LASSO method to construct sentiment index which reducing the complexity of the model while improving the accuracy of the comprehensive index.The LASSO algorithm can reduce the number of variables to reduce the complexity of the model and ensure the explanatory power of the model.The empirical comparison shows that compared with the PCA and PLS methods,the LASSO based regression index has higher accuracy and can accurately measure the investor sentiment in Chinese stock market.(3)Using Quantile regression method,this dissertation empirically tests the impactof investor sentiment on stock market at different loci,and comparing the differences of investor sentiment on the Sino US market.This dissertation adopts both the ordinary least square method and Quantile regression method to test the effects of investor sentiment on stock market returns under the binary market structure.We not only take into account the mutual influence of the distribution centers,but also fully exploit the information contained in the upper tail distribution or the lower tail distribution of the variables.Compared with the US stock market,Chinese investor sentiment has a more significant impact on the volatility of stock market returns.