节点文献

群体模型下的金融市场和资产定价研究

【作者】 唐毅南

【导师】 姜波克;

【作者基本信息】 复旦大学 , 国际金融, 2009, 博士

【副题名】S型需求曲线和金融危机的群体动力学,以及含高阶矩风险升水的期权定价

【摘要】 金融市场是由多种多样的人组成的,但交易者行为加总的宏观结果却有两种截然相反的理论,即代表者模型和群体模型。放任自流的自由资本主义的理论基础是市场出清的代表者模型,即供给和需求相交且仅相交于一点,并且这一点是稳定的。代表者模型假设人的活动由于套利机制相互抵消,金融市场服从布朗运动或随机游走。其漂移项和扩散项没有关联,技术上来讲就可以套利完备,这就是有效市场假说(EMH)(Fama,1970,1991)的最终保障机制,Black-Scholes(BS)模型可以成立(Hull,2000),巨幅波动需以另一单独分布引入(Frisch,1933)。这是和观察到的事实不相符的。由于绝不可能在线性稳定系统中产生出比噪声源更大的波动,EMH加外生冲击的噪声驱动模型(Frisch,1933)不能解释历次金融危机(Chen,2002)。Marshall(1920)最早指出存在不稳定的多均衡,人的行为加总的结果并非总是新古典假设的市场出清。行为金融学则强调人的行为和心理会对宏观系统的行为具有决定性影响,并认为金融危机来自市场的内生机制。De Long和Shleifer(1990)指出价格的暴涨暴跌来自正反馈交易者。但同BAPM(Shefrin,Statman,1994)等模型一样,行为金融学模型中难以测量的参数太多,使得这些工作很难在现实中运用。类似的,凯恩斯主义的Minsky(1975,1982)危机理论缺乏数据分析证据。另外,群体行为的描述在不同的模型中分属一阶(趋势)描述或二阶(波动)描述而没有统一。要在金融学研究中取得进步,首先要避免一个两难问题:收敛的单均衡无法容纳人的行为多样性,而人行为的多样性过于复杂,任何对单均衡的偏离其参数都是很难观察的。本文考虑的是,这个两难可以通过观察可以区分群体模型和代表者模型正确性的经验证据入手,尝试兼顾新古典和行为金融学的优点,用可观测量将金融市场中的群体行为尽量描述清楚。实际上,通过对系统动态的观测来推导微观基础是可能的。例如:Einstein(1905)从布朗运动确定了分子的存在并确立了分子运动论。同样,Chen(2002)发现,宏观指数和股市指数的相对偏差是随时间稳定的,可以用作建立动态随机过程的基准,例如群体运动的生灭过程的相对偏差就趋于常数。李华俊(2002),陈平(2004)和Chen(2005)进一步发现:代表者模型使用的布朗运动和随机游走的相对偏差分别是绝对爆炸和绝对收敛的。从数学上的方便考虑,生灭过程已引入股价运动的描述(Cox and Ross 1976,Kou and Kou 2002)。曾伟和陈平用生灭过程描写两类交易者来修改期权定价模型,以解释波动率微笑的现象,和统一解释已有的期权定价修正模型(2008)。本文进一步考虑非线性生灭过程的高阶矩效益和金融危机之间的联系。本文首先讨论的群体行为的微观机制,Becker(1991)猜测了来自于社会中的群体相互作用的S形需求曲线,该曲线不是单调的并且会产生跳跃。这和Minsky,De Long等的反馈机制是一致的。但是Becker的S形曲线是猜测出来的,缺乏理论,并且和事实有若干不相符之处。本文首先为S形需求曲线建立理论,作为金融市场中群体行为的微观基础。时尚品消费是社会心理的典型现象,其需求变化的特点是跃变而非渐变。本文用人际相互作用下的“社会人”理性构造动态优化模型,来理解S形需求曲线的起源(Becker 1991)。时尚市场的特点是开放型的“进入或退出”机制。消费者心态的转变概率取决于社会风潮的吸引和过度拥挤的厌恶之间的竞争制衡。由此得出的S形曲线的不连续性导致需求的跃变。当社会相互影响强度变弱时,S形需求曲线中不稳定的正斜率部分会退化消失,成为近于单调向下的直线。在理论上,本文没有直接假设金融市场服从特定的随机过程,而是直接观察价格时间序列的转移概率,发现转移概率依赖于系统的状态(即当期价格)。此现象是群体行为的宏观表征,可称为空间非线性。用随机过程通用的主方程处理空间非线性,可以自然得到一个非线性生灭过程。其中,可以将群体行为对所有各阶(趋势,波动,高阶矩)的影响统一描述,相当于将风险从方差推广到包括了偏度(Skewness)和峰度(Kurtosis)在内的更高阶矩,并能直接导出符合经验观察的金融危机的动力学特征——趋势瓦解和高阶矩发散。经验观察发现,这些特征在金融危机时是普遍性的。群体行为产生的高阶矩可以作为金融危机的动力学判据,说明金融危机是内生的。在存在高阶矩的随机过程中,趋势如波动一样是群体行为的结果。因此将过程截断到二阶(方差)并应用伊藤引理也不能做到完全套利,Black-Scholes模型不能很好的描写市场。当高阶矩较小时,使用准线性随机过程逼近真实的价格过程,可以描述高阶矩风险升水,并由此得出可以得出无参数的一般期权定价模型。说明只有群体行为的生灭过程可以同时容纳长时间的市场稳定和短时间的金融危机两种现象。用一般期权定价模型解决BS模型的极值之谜,不但说明BS模型是一般模型的一个特例,也能够间接观测被行为金融学广泛引入但一直难以观测的人的行为参数。

【Abstract】 The financial market consists of various persons.However,there are twocontradicting theories of the macro effect of the micro behaviors:the representativemodel and the collective model.The representative model is the basic of the freecapitalism and neo-classic economy theory,where the demand curve and the supplycurve have one and only one stable point of intersection.The mechanism of therepresentative model is the complete arbitrage,henceforth the efficient markethypothesis (EMH) can be tenable (Fama 1976,1991) and the Black-Scholes (BS)model is right(Hell,2000).In an efficient market,the bubbles and crashes must bespecifically introduced (Frisch,1933).However,the representative model doesn’tallow a jump bigger than the noise source,and the financial crisis has no reasonableexplanation in the EMH framework (Chen,2002).On the contrary,Marshall (1920) firstly noticed the unstable multi-equilibrium,and the collected behavior is not a market clearing.The behavioral finance theoryemphasizes the crucial influence of the human psychology and behavior to the market,and the financial crisis is endogenous (De Long,Shleifer,1990).However,togetherwith other models such as BAPM (Shefiin,Statman,1994),behavioral finance hastoo many invisible parameters.Similarly,the Keynesian theory of crisis (Minsky1975,1982) lacks numerical evidences.In fact,it is possible to deduce the micro foundation via macro observationwithout parameters.Chen(2002) found that the collective process,birth-death process,is the better description of the financial market than Brownian motion and randomwalk.Li (2002) and Chen(2004,2005) also gives the empirical evidences.This thesis begins with the micro mechanism of collective behavior.Beckerguessed an S-shape demand curve in 1991,which can meet the dynamics of thefeed-back of Minsky and De Long theories.Unfortunately,Becker’s S-shape curvelacks theory and cannot meet some important facts.Therefore this thesis tries to buildtheory for the S-shape demand curve first.Sudden changes in demand for fashiongoods are a typical phenomenon in social psychology.Fashion behavior can becharacterized as an‘in-or-out’open market.A dynamic optimization model isintroduced for modeling social rationality where social interactions among populationmembers would change people’s attitudes toward fashion goods.Consumer’stransition probability depends on competition between neighbor attraction and over-crowding.The resulted S-shape demand curve (Becker 1991) is discontinuous atthe positive-sloped part.When the intensity of social influence is weak,the S-shapedemand curve will be degenerated into a negative-sloped normal demand curve.Theoretically,this thesis doesn’t make any presumption that the market follows aparticular model.The direct observation shows that the behavior of the marketdepends on the market state,which means that the collective model is right.Using ageneral master equation of stochastic process,the observation leads to a collectiveprocess:the birth-death process.This thesis then develops the non-linear dynamics of collective behavior and theempirical description of the collective behavior with high-moment premium.Thesetechniques are available in at least two scenarios.First,in this non-linear model,all the moments are collective results of traders,and arbitrage can’t separate trend from the“noise”.Therefore the Black-Scholes (BS)model has a bias about the assumed separation ability.This bias has a smallimperfection in a peace time,and can disable the model in a volatile era.This thesisalso finds that history,expectation,and power do matter in the financial market.Ageneral optionpricing model is then developed,so as to pricing options better than theBS model by resolving the Mythofthe BS Extreme Pricing.Second,financial crisis can be seen as an endogenous phenomenon.Theoreticallyand empirically,this thesis shows that the trend collapse and the high-moment divergeare the dynamic criteria of financial crisis.In another word,it isn’t the noise of Frisch,but the collective behavior driving the mechanism of the financial crisis.

  • 【网络出版投稿人】 复旦大学
  • 【网络出版年期】2009年 12期
  • 【分类号】F224;F830.9
  • 【被引频次】2
  • 【下载频次】1311
节点文献中: