Study On Mean-variance Relation Models Based On Investor Sentiment

Research on factors influencing the return and risk in financial market has been a subject undergoing intense study. During the recent years, with the development of Behavioral Finance, much research has presented new psychology-based theories to solve financial problems. Behavioral Financial Theory emphasizes the aspects of investors’ psychology or emotion in decision-making and market pricing. It explains effectively various anomalies and solves some financial difficulties. However, the research cannot be confined to qualitative description and historical observation. Logical deductions and statistical models are therefore required to describe quantitatively how investor sentiment influences the market.Markowitz used expectation and standard deviation to measure the return and risk.This return-risk relationship is also referred to mean-variance relationship. Due to the fact that the return-risk relationship is time-varying in nature in financial market, this paper adopts statistical methods to construct the revealed models(mainly mean-variance relation models) through the perspective of “investor sentiment”. The main contents of this paper include the following three aspects:Firstly, we conduct an empirical study on the relationship between investor sentiment and performance by applying structural equation models(SEM) from the perspective of the psychological bias and preference that are determined by investor sentiment.SEM with interactions and covariates and multi-group SEM are also incorporated into the study. Based on the survey data, our study finds that investor sentiment including their interactions has a significant impact on investment performance which varies according to individual characteristics( e.g. gender, education etc). In addition, different types of investor show different influential patterns of investment performance.Secondly, we establish a linear varying coefficient ARCH-M model with a latent variable, which is motivated by the psychological factor of time-varying mean-variance relationship. Due to the unobservable property of the latent variable, a corrected likelihood method is employed for parametric estimation. Estimators are proved to be consistent and asymptotically normal under certain regularity conditions. A simple test statistic is also proposed for testing latent variable effect. Simulation results confirm that the pro-posed estimators and test perform well. The model is further applied to examine whether the return-risk relationship depends on investor sentiment.Thirdly, we adopt a threshold model with GARCH effect to study the effect of investor sentiment on the return-risk relationship. The empirical results show that the influence of investor sentiment on return-risk relationship exits a transition mechanism. The relationship is significantly negative when sentiment falls into a low period, is significantly positive when sentiment enters a high period, and is non-significant during the recovery period of market sentiment. In addition, the effect can influence the next market.Furthermore, a change in market policy causes the sentiment fluctuations.