复杂网络视角下的股票市场消息传递行为及其对资产定价影响研究

In recent years, channels of information transmission have substantially been enriched, which essentially changes the networks of information dissemination and the behaviors of message transmission between individuals in stock markets, and makes both of them more complex. The existing literatures investigated the impacts of these changes on asset pricing from two respective perspectives of information networks and message transmission behaviors. However, when sending and receiving messages in real stock markets, individuals always take into consideration such network factors as occasions on which they send messages, and information channels through which they receive messages, and behave differently in the light of these factors. Therefore further researches should be carried out to take these two combined perspectives as a whole and model the information transmission stage more precisely.. This project plans to build agent-based computational models on high performance computing platforms, and to investigate the stage that individuals send and receive messages in complex information dissemination networks and its impacts on asset pricing in stock markets, based on which mechanisms to improve stock market efficiency and to prevent manipulations shall also be designed. Meanwhile, empirical examinations using Internet text mining results and stock market data are also planned to calibrate the aforementioned agent-based computational models and verify the findings and theories. The aim of this project is to deepen the understanding of information dissemination stage in stock market and its impacts on asset pricing, and to provide guidance for regulatory bodies to redesign anti-manipulation rules in the Internet era.

近年来，不断丰富的信息传播渠道使得股票市场中信息扩散网络结构与个体间消息传递行为都发生了根本性变化，并变得日益复杂。现有文献分别从信息网络结构和消息传递行为这两个侧面分析了前述变化及其对资产定价的影响。但真实股票市场中，个体在传递消息时总是会考虑诸如其发布消息的场合、消息来源渠道等信息网络结构方面的因素，而表现出不同的行为。因此应将上述两个"交织在一起的侧面"纳入同一个分析框架开展研究。.　　本项目拟在高性能计算平台上，构建计算实验金融模型，研究复杂信息扩散网络中个体间消息传递行为的动态特征及其对资产定价的影响，并设计可改善股票市场信息效率、有效防范市场操纵的监管机制；同时将结合互联网数据挖掘与真实市场交易数据进行实证检验，校准计算实验金融模型与验证理论。以期更加准确地刻画与认识股票市场信息扩散过程对资产定价的影响，为互联网时代背景下的股票市场反操纵制度设计及其监管实践提供指引。

本项目按照预定计划执行。分别采用计算实验金融、经验实证和理论建模的研究方法，对金融市场中的复杂信息网络动态及其对资产定价的影响开展研究。主要完成了如下3方面工作：.（1）构建了不同市场条件下的金融信息网络之计算实验模型，并相应的研究了其中的金融市场动态规律，这是国际上最早一批提出金融系统信息网络研究框架的工作，并在此基础上创新性的发现了金融系统信息网络之于资产定价的一系列影响。.（2）采用微博数据、股票论坛数据、百度搜索数据、公司股东交叉持股数据、分析师意见等数据非传统数据，开展了资本市场信息网络的动态研究，建立了相应的研究基本范式、框架和研究方法，并取得了一系列的具体领域知识发现。.（3）在上述发现的基础上，试探性通过理论建模，给出了一个投资者异质情况下的资产定价模型。.本项目研究目前已经发表和录用研究论文16篇（其中有5篇论文，因为未及时准确了解标注规定，未能正确标注资助。但这些论文项目负责人都是论文的第一作者或者通讯作者，也确系本项目研究的成果），全部论文均为SCI、SSCI检索的论文。.人才培养方面，在项目研究期中，项目负责人获中组部青年拔尖人才称号，3位博士生参加人毕业赴985高校管理、金融学科任教，其中一人已晋升副教授。