A Study Of The Behaviors Of The Chinese Mutual Fund Investors

China’s mutual fund market has been undergoing high-speed growth since 2013,where the annualized growth rate is as high as 30%.This rapid development continues.By the end of 2021,the asset under management of China’s public funds has exceeded 25 trillion yuan,making it the fourth-largest market in the world.The huge scale and the sustained trend of rapid development make how to better create value for fund investors an increasingly important issue,which has an important impact on the high-quality development of the mutual fund industry.In addition,the rapid development and application of financial technology,and the gradual entry of the post-90s into the labor market,have greatly enhanced residents’ financial awareness.This trend and the accumulation of household wealth have led to the result that financial assets,especially mutual funds,account for a much more significant proportion of household wealth allocation.This makes the aforementioned question of how to better create value for investors related to the interests of the fund industry and investors and directly contributes to major national strategic issues such as common prosperity.To develop and adopt appropriate policies to effectively protect investors’ interests,an in-depth understanding of fund investors’ behavior is necessary.Theoretically,investors should learn the fund’s skills from historical fund performance based on Bayesian learning to form rational expectations of the fund’s skills and make investment decisions accordingly.However,recent studies show that investors do not always follow Bayesian learning to evaluate fund performance.Moreover,the Chinese fund market has a number of distinctive features that might make Chinese fund investors’ behaviors deviate even more from the aforementioned theoretical implications.For example,the Chinese mutual fund market has a relatively short history and has experienced rapid growth over the past few years,making it one of the most important emerging markets.Thus,the Chinese market is different from the well-studied US fund market.Moreover,most Chinese fund investors are individual investors,who are less sophisticated and may be more severely affected by constraints such as limited attention and limited endowment.The above characteristics of the Chinese fund market,especially the constraints faced by investors,also make the behaviors of Chinese fund investors unique.In this study,I aim to provide a deeper understanding of Chinese fund investors’ behaviors and the mechanisms that lead to such behaviors.Thus,I focus on three sets of typical facts related to Chinese fund investor behaviors as follows.First,this paper finds that,given the same fund skills/performance,Chinese investors respond more strongly to funds that appear to consistently earn profit,i.e.,funds that make the investors has a stronger perception of performance persistence.However,these funds with a stronger perception of performance persistence for investors do not deliver better returns.Second,we find that Chinese fund investors respond significantly not only to the fund skills(alpha),but also to factor-related returns(FRR)and market-related returns,and that investors’ responses to fund FRR depend on market states and are significant only in the moderate markets.This behavior is quite different from the implication that "rational investors should focus only on fund alpha" as suggested by existing studies.Finally,drawing on psychological research,this paper finds that investor experience has a significant impact on the fund investor behavior.While the existing studies show that investors care about simple signals such as historical fund returns and sophisticated signals such as alpha.I find that investors also respond significantly to the model-free expectations of a fund’s skills,a signal that portrays the simple learning of fund investors based on experience.To understand the mechanisms underlying the above phenomena,I conduct an exhaustive theoretical and empirical analysis in this study.The analysis shows that the constraints faced by investors have a significant impact on investor behaviors.Specifically,investors’ limited attention implies that investors can process only a limited amount of information during a given period.This constraint leads investors to make mistakes in learning the fund’s skills and thus react more strongly to funds with a stronger perception of performance persistence.The limited endowment constraint makes it difficult for investors to effectively hedge a fund’s systematic risk,which in turn leads to the finding that investors respond significantly to different components of fund excess returns,i.e.,alpha,factor-related returns,and market-related return.The limited learning ability is a special case of limited attention.This constraint directly makes investors unable to engage in rational Bayesian learning.To learn the fund’s skills,fund investors with limited learning ability can only turn to model-free learning approaches that require little learning ability.On this basis,we can draw clear policy implications.It has been commonly argued that Chinese fund investors,who are predominantly individual investors,are not sophisticated enough and are vulnerable to behavioral biases.Thus,the existing studies mainly attributes investor behaviors that deviate from rational Bayesian learning to investor behavioral biases and accordingly suggest to focus on investor education.The research in this paper supports this argument by showing that accurate investor education is indeed important to protect the interests of fund investors.However,the paper also suggests that investor education should be conducted according to the investor sophistication(e.g.,to ameliorate the effects of investors’ limited learning ability).More importantly,the research in this study also indicates that investor education alone may not be sufficient.For example,to change investors’ significant response to funds’ factorrelated returns induced by the limited endowment constraint,it is more important to strengthen the financial markets and provide more effective hedging instruments available to investors constrained by the limited endowment.This paper consists of six chapters.Chapter 1 is the introduction,which carefully explains the research background of this paper,the research ideas,methodology and structure of the paper,as well as the innovations and shortcomings of this paper.The second chapter is a literature review,which carefully summrizes the literature on fund flows,the analysis of economic consequences of investors’ behavior,and behavioral finance that are closely related to the study of this paper.Chapters 3 to 5 then examine the effects and mechanisms of investors’ limited attention,endowment constraints,and investors’ limited learning ability on fund investor behaviors,respectively.Chapter 6 concludes the study by summarizing the research of this paper and discussing further research directions in related areas.Specifically,Chapter 3 explains why investors respond more significantly to funds with a stronger perception of performance persistence based on the investor limited attention.Fund investors can extract a return signal from a fund’s historical performance to learn the fund’s skills.Among other things,signal precision has an important impact on investor learning.The existing studies usually assume that the signal precision is known,but the non-normal characteristics of fund performance and the time-varying nature of fund skills make this assumption untrue in reality.Therefore,investors need to estimate the signal precision.Investors constrained by the limited attention can only use simple methods to estimate the signal precision.I find that investors estimate the signal precision based on the perceived return persistence(PRP)transmitted to investors by the fund.This makes investors overestimate the signal precision of funds with a high PRP and overreact to them,leading to a mismatch between the fund scale and skills.In other words,investors of these funds will suffer additional losses in the future.Chapter 4 further examines how the limited endowment constraint affects investor learning and makes investors’ responses to fund factor-related returns dependent on the market states.The limited endowment constraint implies that investors have only limited investable assets,thus making them unable to effectively hedge funds’ systematic risk.This makes investors focus on the fund excess returns.Since fund excess returns can be decomposed into alpha,factorrelated returns(FRR),and market-related returns,the limited endowment constraint then makes investors respond significantly to all three of these components.Further,when the average realized factor return over the past periods is not too extreme(in a moderate market),the fund’s factor-related returns is persistent and thus have significant positive predictive power for fund excess returns.Thus,investors who care about the fund excess returns will rationally respond significantly to the factor-related returns.Conversely,when realized factor returns over the past periods are extreme(in the volatile market),the fund’s factorrelated returns do not help predict the fund’s excess returns and investors no longer respond significantly to them.In contrast,investors’ response to the fund alpha is always significantly positive and is independent of market states.The above findings indicate that the behaviors of Chinese fund investors are different from those in developed markets such as the US: the response of US fund investors to the fund alpha varies with the state of the market(Franzoni and Schmalz,2017).Chapter 5 examines how fund investors subject to the limited learning ability learn the fund’s skills from a model-free learning perspective.In this chapter,I refer to "limited learning ability" as a special form of limited attention,where the amount of information that an investor can process is less than what is required to conduct Bayesian learning,making it impossible for the investors to perform rational Bayesian learning.I argue that these investors can learn the fund’s skills via a simple learning approach named the Q-learning,a model-free learning method that requires little learning ability and little prior knowledge from the investor.To this end,I extend the framework of Barberis and Jin(2021)on modelfree learning for investors,and construct a binomial tree for each fund.Then I propose a model-free expectation indicator(MFE)of the fund’s skills to characterize the expectations of the fund’s skills for investors who adopt the model-free learning approach,and to examine its impact on fund investor behaviors.I find that MFE has a significantly positive impact on fund flows.MFE also has a significantly positive predictive power for future fund performance,suggesting it can be a viable alternative to the more specialized alpha for learning the fund’s skills.Chapter 6 concludes the paper.The core contributions of this paper include the following three aspects.First,the paper shows that investor behaviors that deviate from rational Bayesian learning are the results of constraints faced by investors,and that different constraints have different connotations.Accordingly,in order to better protect investors’ interests,it is necessary to first understand the mechanisms underlying investors’ behavior and then make appropriate policy choices in a targeted manner.Second,this paper contributes to the hot debates about "investor sophistication" in recent years.The existing studies on fund investor behaviors hold extreme and opposing views,either that investors are highly sophisticated or that they are fully unsophisticated.This study shows that the fund investor behavior is not always extreme,and that even in the face of various constraints,investors still attempt to learn the fund’s skills,via simple approaches such as Qlearning or selectively responding to the fund’s factor-related returns.Finally,from a methodological perspective,this paper provides the first empirical analysis of model-free learning in financial markets,arguing that investors do adopt modelfree learning as a simple learning approach.This study is also the first paper to study the model-free learning of mutual fund investors.