【作者】 张欢；

【导师】 杨昊晰；

【作者基本信息】 南开大学 ， 金融工程， 2018， 硕士

【副题名】基于中美市场典型特征因子的实证分析

【摘要】 第五次全国金融工作会议指出,防止发生系统性金融风险是金融工作的永恒主题,而如何主动化解市场极端风险是实现这一主题必不可少的环节。如何测度市场极端性事件的发生概率已成为近年来学术研究关注的热点。现有研究表明各金融资产的收益率的尾部相依性可以准确的刻画金融资产间存在的非线性相依关系,为准确测度市场极端性系统性风险提供了理论依据。与此同时,实证证据进一步指出金融资产收益率在市场处于下跌尤其是暴跌状态时,各资产收益率之间存在更显著的联动性。由此可见,金融资产收益率间的尾部相依存度在不同市场状态下程度不同,这一特征又被成为尾部相依结构的非对称性。因此,本文旨在对中国、美国股票市场各类权益类资产组合与市场之间的尾部相依结构及其非对称性进行比较分析。明细市场极端性风险的传导路径与发生根源,从而为早发现、早防范极端性系统性金融风险的发生提供了理论依据与实证证据。本文选取1996年7月1日至2017年12月31日的中国股票市场日度数据,分别构造了基于规模因子、账面市值比因子、动量因子以及流动性因子的股票资产组合,并测算各资产组合收益率与市场组合收益率分布的尾部相依性。为了进一步探究我国股票市场与较为成熟的美国股票市场之间的异同性,文章同时选用了1966年7月1日至2017年12月31日美国股票市场日度数据与中国市场进行对比分析。在度量各个特征因子股票组合与市场组合收益率之间的尾部相依性时,文章分别采用了超位相关函数(EC)和尾部相依系数(TDC)两种方法。为避免时间依赖性对于收益率分布结构的影响,本文引入AR(p)-GJRGARCH(1,1)过程对收益率序列的一二阶矩的自相关性进行修正。采用Hill统计量对市场组合和各个特征因子股票组合收益率的尾部指数进行计算度量,结果表明过滤后各个收益率序列的高阶矩都是稳定存在的。文章的实证结果表明,首先,中美市场上各个特征因子股票组合尾部相依性均存在非对称性,且左尾相依程度普遍强于右尾。与此同时,美国股票市场数据呈现的非对称程度要强于中国市场。此外,在中国市场,各个股票组合与市场组合收益率之间尾部相依程度随着规模因子的升高而增强,同时随着账面市值比因子、动量因子以及流动性因子的升高先增强后减弱。与中国市场所得到结果不同,在美国市场上,各个股票组合与市场组合收益率之间尾部相依程度随着规模因子的升高而增强,随着账面市值比因子的升高而减弱,随着动量因子的升高先增强后减弱,随着流动性的增强而减弱。研究结果还指出,两市场尾部相依结构的非对称性普遍与尾部相依程度走势相反。文章随后进一步对尾部相依性的度量方法进行讨论,结果指出超位相关函数较尾部相依系数对于非对称性更为敏感,而过滤掉时间依赖性对于尾部相依系数的影响非常小,而对超位相关函数来说则可以增加其区分度。结合所得研究结果,本文最后对于我国股票市场的审慎性监管模式提出了几点思考,指出监管者应重点关注规模较小、账面市值较低、历史表现较差(即动量因子较低)以及流动性过弱或过强的股票组合。本研究旨在为我国进一步完善证券市场系统性风险监管体系提出方法上的修正,并以此为目标构建早预警,早应对的全面监管体系。更多还原

【Abstract】 The Fifth National Conference on Financial Works emphasizes that,preventing the occurrence of systemic risk events is the essential task of the financial work in China.Therefore,how to actively avoid the occurrence of extreme risk events is the key point.Recent research tends to investigate that methodologies which are capable to measure the probability of occurrence for the rare extreme risk events.Existing theoretical evidences show that tail dependence of financial asset returns is able to capture the non-linear dependence relation among assets.Meanwhile,empirical studies reveal the degree of tail dependence would be enhanced during crash periods.Hence,within different scenarios,the tail dependences between financial assets perform asymmetrically.Therefore,this thesis aims to investigate the asymmetry of tail dependence between different anomaly-based portfolios and stock market returns for both China and US and provides both theoretical and empirical evidence for the systemic financial risk regulating.Based on the daily individual stock data from July 1^st,1996 to December 31^st,2017 of Chinese markets,this paper constructed size,book-to-market,momentum and liquidity portfolios and calculates the tail dependence measure between each portfolio return and market return.To be more precise,this paper also use the daily data set of US market from July 1^st,1966 to December 31^st,2017 to do the same exercise and compare with the results from Chinese market.I mainly use the extreme correlation（EC）and the tail dependence coefficient（TDC）methods to construct the tail-dependence structures and measure the asymmetry properties.In order to avoid time dependence effect on the distribution of asset returns,this paper introduces the AR（p）-GJR-GARCH（1,1）model to filter the autocorrelation of the first and second moments of the returns.In order to ensure the existence of nonlinear dependent structures,this paper verifies the existence of the corresponding market anomalies,and uses Hill statistics to measure the tail index of each portfolio.The results finds that the high-order moments of the return series consistent exist after filtering,which brings the necessary environment for measuring the tail dependence structure.This paper finds that the tail dependence of each portfolios on both Chinese and US markets shows asymmetry,and the degree of the left tail dependence is generally stronger than that of the right part.Then,the asymmetry of tail dependence in the US market is stronger than that in the Chinese market.Specifically,in the Chinese market,the degree of tail dependence increases with the raise of the size,but increases first and then decrease with the enlarge of the book-to-market ratio,momentum and liquidity.Moreover,in the U.S.market,the degree of tail-dependence increases with the increase of the size,decreases with the increase of the book-to-market ratio,increases first and then decreases with the increase of the momentum factor,and decreases with the increase of liquidity,while the trend of asymmetry is also opposite to that of the tail dependence.This article then compares and discusses the measurement of tail-dependency.The extreme correlation is more sensitive to asymmetry than the tail dependence coefficient,while filtering out the time dependence has very little effect on the tail dependence coefficient.However,there is huge effect on the extreme correlation by increasing the differentiation.Finally,based on the analysis results,this paper provides some thoughts on the financial regulating process for the Chinese stock market.The regulatory department should focus on stock portfolios that are small in size,have a low book value,have poor historical performance（i.e.,a low momentum）,and have too weak or excessive liquidity.Aiming to provide a more comprehensive and systematic risk measurement method for the securities market,there is also discussion on the direction of further research.更多还原