【作者】 李凯；

【导师】 王哲伟；

【作者基本信息】 山东大学 ， 西方经济学， 2020， 硕士

【摘要】 互联网逐渐成为人们获取资讯的主流渠道。股吧、雪球网等互联网投资类论坛上汇聚了几百万投资者,他们之间形成了一个信息交流的庞大复杂社会网络,并且会反过来影响每一个人的预期形成和投资决策过程。基于互联网投资类论坛所形成的庞大社会网络,对微观个体来说,能否增加个人私人信息并且提高投资收益?对宏观市场而言,这种去中心化的复杂社会网络上信息交流,是否能够提高市场定价效率?这是本文研究的核心问题。在梳理有关文献及理论的基础之上,本文使用Python编写爬虫程序进行了数据准备。具体来说,从雪球网这一社交投资平台获取了 36万用户1000多万条互相关注的信息,通过Gephi绘制了这些用户之间的社会网络;爬取雪球网上70多万虚拟投资组合的持仓及历史收益情况,据此形成了由社会网络到投资收益的完整映射。实证方面,本文从微观层面和宏观角度两个层面对投资者论坛与资产定价之间的关系进行了研究:第一、从微观层面出发,从投资者个体私人信息获取角度研究了互联网投资类论坛对个人投资收益的影响。基于有效市场假说及信息不对称理论,从理论上分析了论坛的积极作用和消极作用。为了描述社会网络上不同节点之间的相互影响,本文引入了图论的“输出度”概念测度信息来源的数量,引入“中心性”测度社会网络中不同节点的重要性差异,并将一个节点邻居节点的平均中心性定义为“信息质量”。研究结果表明:(1)投资者邻居节点的平均中心性越高,投资者获得的信息越多;(2)投资者的“输出度”(信息来源的数量)对信息拥有量的影响不显著,说明盲目关注更多用户并不会给用户带来信息增益;(3)投资者信息质量和信息数量提高了均提高了个人收益水平;(4)分位数回归结果表明,社会网络对投资者的信息增益和收益提升作用存在异质性,收益水平越低的群体,从社会网络中获益的程度越高。第二、从宏观层面,研究互联网投资类论坛对资产收益率的影响。本文使用CNRDS-GUBA数据库的情绪分类结果,构建情绪指数和意见一致指数,研究发现:(1)情绪指数与资产收益率之间存在正相关的关系,这与以往学者的研究结果一致。(2)使用DCC-GARCH模型研究了情绪指数与资产收益率波动关系的动态变化,研究结果表明:发帖量与相关性正相关,发帖量越高,相关性越高。第三、再次从宏观层面,研究互联网投资类论坛对资产定价效率的影响。首先基于理论层面的分析提出研究假设,认为互联网投资类论坛能够提高市场定价效率。其次,参考以往学者对资本市场定价效率的界定,基于信息熵理论和股价同步性分别构建定价效率指标efft_entropy和RSQ;最后,通过面板数据的实证研究发现,互联网投资类论坛上的交流行为(发帖、阅读、评论)均促进了当期及下一期资产定价效率的提高。本文的主要贡献在于:第一、构建了一个包括36万互联网用户的大型复杂社会网络,并实证研究了其对投资者个人的影响。第二、从微观角度出发,研究互联网投资类论坛对个体投资者的影响,而在以往学者的研究当中,大多数是研究互联网投资类论坛对资本市场的影响,微观视角的研究较少。第三、引入信息熵这一概念,构建熵有序度指标用于衡量微观个体的信息能力,以及用于衡量股票市场的无序程度和定价效率。通过以上研究,本文认为,互联网投资类论坛,在微观上能够起到帮助个体提高收益,这可能是因为个体从论坛获得了信息;在宏观上,互联网投资类论坛所表现出来的投资者情绪与市场收益率之间存在着动态的相关关系;论坛上广泛的讨论行为(发帖、阅读、评论),也提高了股市的信息含量和市场定价效率。更多还原

【Abstract】 The Internet is becoming the main way for people to get information.There are millions of investors have gathered on Social Trading Platforms,such as Xueqiu and Guba.They have formed a huge and complex social network for information exchanging,which will in turn affect the formation of human expectations and the investment decision process.Based on the huge social network,can the Social Trading Platforms increase their personal information and increase investment returns?For the asset pricing,can the platforms improve the market pricing efficiency?This is the core issue of this paper.On the basis of relevant literature and theories,this paper used Python to write a crawler program for data preparation.We obtained more than 360,000 users’ 10 million pieces of mutual attention information from xueqiu.com,then mapped the social network between these users by Gephi;We also crawled more than 700,000 virtual portfolios on Xueqiu to know users’ returns.The relationship between Social Trading Platforms and asset pricing was studied from a micro perspective and a macro perspective:First,from a micro level,the impact of Social Trading Platform on individual investment return was studied.Based on the efficient market hypothesis and the information asymmetry theory,the positive and negative effects of the platforms were theoretically analyzed.In order to describe the interaction between different nodes,this paper introduced the concept of "out degree"(source of information)and "centrality"to measure the difference in importance of different nodes in the social network,and The average centrality of a node’s neighbors’was defined as "information quality".The research results show that:(1)Investors obtained information from neighbor nodes on the social network.The higher the average centrality of neighbor nodes,the more information investors get;(2)The "out degree" of investors had no significant effect on the amount of information,indicating that blindly paying attention to more users would not bring information gain to the users;(3)The increase in information quality and the amount of "out degree" had increased the level of personal returns;(4)The quantile regression results showed that the social network has heterogeneity in investors’information gain and income enhancement.The lower the return level,the higher the degree of benefit from the social network.Second,from a macro perspective,the impact of Social Trading Platform on asset returns was studied.This paper used the sentiment datas in CNRDS-GUBA database to construct the sentiment index and opinion divergence index.The study found that:(1)There was a positive correlation between the sentiment index and the asset return,and there was a negative relationship between the opinion divergence index and the return rate.(2)The DCC-GARCH model was used to study the dynamic change of the relationship between the sentiment index and asset returns.The research results showed that the postings volume was positively correlated with the correlation.Third,from a macro perspective,the impact of Social Trading Platform on asset pricing efficiency was studied.Based on the theoretical analysis,the Social Trading Platforms was thought improving the asset pricing efficiency.Referring to the previous scholars’definition of capital asset pricing efficiency,based on information entropy theory and stock price synchronicity,this paper difined efft_entropy and RSQ to measure pricing efficiency;finally,through empirical research on panel data,we found that the communication behavior on Social Trading Platform(posting,reading,and commenting)had all contributed to the improvement of asset pricing efficiencys.The main contributions of this paper were as follows:Firstly,a large-scale complex social network with 360,000 Internet users was constructed,and its impact on investors was empirically studied.Secondly,from a micro perspective,the impact of ISocial Trading Platform on individual investors was studied.There are few studies on micro perspectives.Thirdly,introduced the concept of information entropy,and defined a new index using entropy to asset pricing efficiency of the stock market.Through the above research,this paper believed that Social Trading Platform can play a important role in helping individual investors to obtain information and improve returns;on the macro level,the discussion activities(posting,reading,commenting)on platfroms had also improved the the efficiency of market pricing.更多还原