【作者】 许敏；

【导师】 刘善存；

【作者基本信息】 北京航空航天大学 ， 管理科学与工程， 2010， 博士

【摘要】 市场状态与交易者行为的相互关系以及信息性交易的准确测度,是指令驱动市场微观结构理论的研究热点和难点。论文分三部分、各两章内容研究市场状态如何影响交易者行为、交易者行为对市场微观特征形成的反作用以及信息性交易的测度。主要研究内容及结论如下:第一、以金融市场微观结构理论为框架,基于成交序列的高频分笔交易数据,实证研究上海证券市场投资者的风险态度。从投资者个人理性出发,基于所有交易者形成的限价指令簿基本特征——买卖价差、深度,探讨二者如何随市场状态的变化而变化;采用LSB价差分解模型,分析交易者行为变化的风险原因。研究表明市场流动性与市场波动性水平的变化对交易者行为影响较大;交易者调整其交易行为以规避信息不对称风险。第二、扩展EKOP模型研究我国交易者的市场到达强度及其影响因素。将交易者市场到达强度看作依赖市场状态的变量,信息结构不同的投资者其市场交易行为具有不完全相同的市场依赖因素和依赖程度,表现出交易强度具有时变特征。实证表明我国市场非知情者主要关注于宏观变量,如市场收益、总成交量、市场波动性,而知情者除了观察这些变量以外,还关注股票收益、平均每笔成交量、相对价差、市场深度、市场弹性。第三、从时间序列的角度研究交易者行为对市场状态——波动性的反作用。将交易量、波动性作为内生变量,久期与交易方向作为外生变量,采用VAR模型研究在不同的市场趋势下,交易量与波动性之间的动态关系,并通过脉冲分析,研究未预期交易对波动性的持续影响。在牛市,有更多的私有信息存在于交易中,单位未预期交易可以引起波动性更大幅度的变化,且信息被市场完全吸收所需的时间更长。第四、从横截面角度研究交易者行为对市场状态——波动性的反作用。按照不同的市值或行业将股票分为若干投资组合,采用组合预期收益离差(ERD)衡量交易者信念的异化程度,研究其是否包含组合或个股收益波动性的信息。在牛市中ERD对组合或个股收益波动性的影响是剧烈的,而熊市中组合或个股收益波动性对ERD的反应并不显著,其主要受组合和指数时间序列条件方差的影响。第五、很多学者采用EKOP模型中的PIN指标作为信息性交易概率代理变量,但只是默认了这种方法的有效性。本文提出扩展EKOP模型,采用两种方法检验传统上采用交易笔数计算的PIN指标是否是信息性交易概率的有效代理。一种是从知情者与非知情者收益的角度,另一种是从公司公告前后信息性交易概率差异的角度。结果表明采用交易量或交易金额加权的交易笔数扩展EKOP模型可以将知情者与非知情者区分开,计算所得PIN指标更符合信息性交易概率的真实情况。第六、运用收益方差分解方法,基于股票市场交易者多类型异质现象,研究不同类型交易者的信息性交易比例。将交易者区分为三种类型:提前知道公告信息的第一类知情者(内幕交易者)、分析判断股票信息的第二类知情者以及非知情者。运用事件研究方法,比较公告信息前后信息性交易比例的差异,发现内幕交易者在市场中占的比例并不大,平均只有2.78%,36%的知情交易者主要靠独立的分析判断进行交易,如各种机构投资者。本论文为交易者如何进行投资决策提供了理论支持,为监管者如何从微观结构角度了解证券市场提供了参考。更多还原

【Abstract】 The interrelationship between market status and investor behaviors, and the measurement of informed trading are two prevailing and difficult topics in limit order driven market. This dissertation consists of three parts, and each part includes two chapters for discussions in details. The paper states the influence of market status on investors’behaviors, the impact of investors’behaviors on market status inversely, and the informed trading measurement. The details are as followed.First, Chapter Two studies investors’risk attitude and risk factors from a micro-level, based on high-frequency trading data. Owning to rational individual, spread and depth represent investors’common behaviors, which are formed by limit order book. This chapter investigates the real reason of investor behaviors change, adopting LSB spread separation model. The results indicate that liquidity and volatility change ratio have more impact on investors’behaviors; the investors mainly have concerns over asymmetry information.Second, Chapter Three discusses the investors’arrival intensity and the influencing factors, adopting EKOP model. The arrival rates are time-variation, and largely depend on different market situation. The uninformed traders focus on some macro variables, including market return, total volume and market volatility. While informed traders have concern on some micro variables, such as stock return, average volume, relative spread, market depth and market elasticity besides of those macro variables.Third, Chapter Four investigates reaction of investor behaviors on volatility, one of market situations by time series. This chapter, which adopts VAR model, studies the relationship of volume and volatility in different market situations, consisting of volume and volatility as endogenous variables, duration and trading direction as exogenous variables. Then it moves further into the persistent impact of unexpected trading on volatility through impulse respond analysis. Comparing to bear market, there is more private information in bull market, one unexpected trading could result in larger change on volatility, and the private information could be absorbed totally by longer time.Fourth, Chapter Five investigates reaction of investor behaviors on volatility, one of market situations by cross-sectional angle. Firstly, all the stocks in Shanghai Stock Exchange are divided into several portfolios based on capitalizations or industries. Then, it measures heterogeneous beliefs of investors using cross-sectional expected return dispersion (ERD). Lastly, it verifies whether ERD includes the new information about volatility. In bull market, expected return dispersion affects volatility of portfolios and individual stock fiercely. However, this kind of effect is not significant in bear market. In bear market, return volatility is decided by time-series conditional variance of portfolios and index.Fifth, PIN index in EKOP model was adopted to investigate the probability of informed trading in many literatures. However, most of these studies did not directly test the veracity. Chapter Six suggests the extensional EKOP model and investigates the validity of PIN from two methods: the returns of informed and uninformed traders; the behaviors of PIN in different periods before and after announcement. The empirical results show that the traditional PIN index computed by numbers of trading is worse than extensional model that computed by volume or value.Sixth, Chapter Seven studies different investors’informed trading proportion with return variance decomposition method to decompose the variance of returns, based on heterogeneous investors. It subdivides investors into three categories: the traders who acquire announcement information in advance (the first type of informed traders, insider trader); the traders who analyze information privately (the second type of informed traders) and the uninformed traders. Comparing the difference of informed trading before and after announcement, it discovers insider trading percentage is small, which accounts for 2.78%. The independent analysts, for example institutional investors, are 36% on average.This dissertation provides theoretical foundation for investors on their trading decisions and presents a valuable access for regulators on understanding security market.更多还原