无忧论文网
当前位置: 无忧论文网 > 社科论文 > 会计审计论文 > 财务会计论文 > 我国上市公司股权分置改革对价决策的行为特征——基于学习理论与羊群理论的分析
点击提交论文指导需求
高薪诚聘老师
我国上市公司股权分置改革对价决策的行为特征——基于学习理论与羊群理论的分析
时间:2011-02-21 浏览次数:1692次 无忧论文网
点击这里在线咨询我
财务学
行为公司财务行为公司财务
    论文摘要   中国上市公司的股权分置改革是一次在中国证券史上具有里程碑意义的重大事件。本次股权分置改革在程序上具有三个影响深远的特点:一是地方政府和各级国资委可能积极地发挥着影响力;二是股权分置改革是从少数公司试点到逐步全面推开;三是由非流通股东先行制定股改方案,然后由流通股东对该方案进行表决,而不是双方股东共同拟定一个股改方案。本次股权分置改革中,地方政府和各级国资委的积极介入可能导致改革结果偏离了股东权益的帕累托最优化;从试点到全面铺开的程序,决定了股改初期试点公司的对价方案出台后,对后续股改公司将会产生巨大的示范效应和压力,这种示范性压力可能影响到非流通股东和流通股东之间的相对谈判实力。由非流通股东先行制定方案,方案出台距离股东大会召开平均有一个月的间隔,在这期间,陆续股改的公司可能参照了先期股改公司的方案而出台自己的方案,新方案的出台反过来又对即将进行表决的先期公司的流通股东产生心理压力。事实证明,股权分置改革的程序决定了非流通股东与流通股东的力量对比悬殊,并且谈判砝码不断向非流通股东倾斜。因此,尽管股权分置改革在形式上表现为非流通股东与流通股东的双方博弈,但实质上,股权分置改革乃是以非流通股东为主体的决策过程,在这一决策过程中,流通股东并未发挥应有的作用,各行为主体之间的互动主要表现为不同公司的非流通股东之间的虚拟博弈过程。因此,我们在虚拟博弈理论的框架下,基于学习理论和羊群行为理论,研究本次股权分置改革对价决策的行为特征。本次股权分置改革过程表现出三个显著的特征:随着股改批次的扩大,对价比例逐渐下降;对价比例具有显著的群聚性;市场在股改期间结束了长达4年的熊市,开创了新一轮牛市。那么学习行为可以解释对价比例的下降趋势吗?所观察到的对价群聚性是学习过程的收敛所引致的,还是羊群行为的结果?为什么股改公司要采取羊群行为?证券市场的表现影响对价决策吗?在股改的流通股东与非流通股东之间,市场态势的变化更有利于谁呢?对价的群聚性与此有关吗?本文通过对股权分置改革程序的分析,建立股权分置改革的虚拟博弈假设;通过对心理学、经济学、财务学有关学习理论和羊群理论的研究文献的梳理,建立了学习理论和羊群理论的比较分析;在此基础上,建立股权分置改革“学习行为假说”和“羊群行为假说”;并以沪深两市全部非ST类股改公司和非ST类采取纯送股方式的股改公司为样本对学习行为(包括最优反应学习和模仿学习)的存在性、表现方式、收敛特征、学习的结果)、羊群行为存在性及其影响因素以及股价表现、股改阶段及参与者数量等内、外部因素与对价行为特征之间的关系等进行了实证研究和对比分析,为股权分置改革“学习行为假说”和“羊群行为假说”提供了证据。从理论上分析,学习是指在人类行为可观测的环境中,通过对个人经历、环境结构以及他人行为或结果的观察、分析、接纳来改进自己的决策信息,从而提高决策的质量或效用。学习的根源来自信息不对称和人的有限理性。羊群行为是指为了提高效用(包括决策正确性、决策收益、声誉或个人薪金等管理者私利),决策者放弃私有信息,转而模仿、跟随前人或众人的决策。羊群行为的根源在于信息不对称、社会规范和代理问题。学习行为与羊群行为的联系与区别表现为:学习理论和羊群理论都研究群体决策与行为,都关注决策者之间的相互影响和环境对决策者的影响;羊群行为部分地始自学习的动机;学习行为的结果表现为羊群现象;社会学习是通过群体互动将私有信念(信息)扩散到所有决策者,它促进了公共信息的增长;而羊群行为是通过群体互动而删除或弱化了私有信息,从而阻断了私有信息向公共信息的渗透,阻碍了公共信息的继续增长。表现在实证特征上,学习的结果使得群体的行为可能收敛,也可能分散,如果收敛,而且学习链足够长,则群体行为收敛于“真值”;而羊群行为的结果必将导致群体行为收敛,但未必收敛于“真值”。在信息机制下,学习的决策质量一定逐步改进,决策效用一定逐步提高,而羊群行为的决策质量降低,群体决策效用降低;在代理机制下,羊群行为的个体效用可能得到提高。通过建立股权分置改革学习行为的理论模型,我们发现,在股权分置改革这个兼具序列性、重复性和复杂性的决策过程中,学习行为既有模仿学习的成分,也有最优反应学习的成份。其中最优反应学习过程既包括强化学习,即通过观察而强化前人策略的吸引力;也包括虚拟博弈过程,即根据观察而修正个人信念。实证研究为理论分析提供了证据。实证研究发现:在股权分置改革过程中,同时存在着最优反应学习和模仿学习,最优反应学习揭示了对价比例下降趋势的内在机制;模仿学习表现为:在时间序列[1]上,对价比例在相邻批次之间表现出粘性,并引致对价在横截面上的(部分)聚集效应。粘性的累计并未影响对价比例的持续下降趋势,信息层叠(Cascades)并未发生。我们的研究还发现,非流通股份所占比例越低、公司效益越好、机构投资者持股比例越高、股改方案所包含的业绩或分红附加承诺越强、股票估值越高(相对于每股收益),非流通股东所支付的对价越低。显然,当市场“蛋糕”做大了之后,市场上涨更有利于非流通股东:非流通股东成功地使对价比例下降,因而他们从市场上涨中得到的收益更大。但机构投资者作为最大的流通股东,不仅没有争取到高的对价补偿,反而发挥了相反的作用——机构投资者持股越多的公司,对价比例反而越低。 我们还从理论和实证两方面研究了股权分置改革中对价决策的羊群行为。通过建立理论模型,我们得到以下主要结论:经济环境、产业竞争、决策者理性程度等因素的变化都会影响独立决策质量,决策者具有采取羊群行为,以寻求更多的信息或心理支持的动机。在群体中,如果存在着产业竞争,并存在资本市场对资产价值的高估,则决策的羊群行为总是存在的。当资产价值被资本市场低估,则决策者易于产生保守行为,未必发生羊群行为。决策者对项目风险的关注以及对项目综合收益的精确预算可能降低发生羊群行为的概率。对股权分置改革羊群行为的实证研究包括时间序列上羊群行为的存在性;横截面上对价的群聚性和羊群行为的存在性;羊群行为条件下的对价影响因素;以及羊群行为的影响因素等。通过实证研究,我们发现:(1)股权分置改革在时间序列上不存在羊群行为,在横截面上,除了存在学习引致效应而产生的对价群聚性外,还存在着对价决策的羊群行为。说明中国上市公司在制定股权分置改革对价决策的过程中,既存在着对前人决策方式的学习行为,也存在着对同一批次其他公司对价决策的羊群行为。除此之外,在对价决策中,决策者还考虑了本公司的盈利能力、非流通股份比例、机构投资者持股比例和股改方案所能做出的附加承诺等特定因素。而公司规模和股价上涨等因素的独立影响不再显著,可能隐含在学习行为或羊群行为之中。(2)股权分置改革的政策性特征限制了决策者的选择余地和独立决策空间,因此无法有效避免对价决策中的羊群行为,也扭曲了决策的羊群行为与资本市场估值之间的关系。(3)盈利能力越强的公司,不仅对价比例越低,而且越倾向于在对价决策中采取羊群行为,说明盈利能力强的公司为了尽可能减少所支付的对价,更加倾向于通过从众来寻求支持,在对价决策中采取羊群行为可有效地增加流通股东的心理压力,减弱流通股东的谈判势力。而盈利能力相对较弱的公司,由于具有更强的风险顾虑,可能在决策中更为审慎。股改前非流通股比例没有显著影响对价决策的羊群行为。这可能因为上市公司非流通股份比例的平均水平较高,股改并不会威胁大股东的控制权。上市公司的负债率水平越高,对价决策的羊群行为程度越低。(4)对价决策的羊群行为与同批次参加股改的公司数量负相关,说明当不存在竞争性且群体规模较大时,羊群行为有所减弱。(5)随着批次的加大,对价的羊群行为越来越强烈,结合“随着批次的加大,对价比例越来越低”的现象,说明采取羊群行为的确具有策略性意义:在对价决策中采取羊群行为可以有效地增加流通股东的心理压力,减弱流通股东的谈判势力,以使越来越低的对价比例能够顺利获得流通股东的同意。随着股改深入,上市公司对羊群行为的效用的认识也不断扩散。总之,股权分置改革过程中既存在着信息不对称的问题,同时,很可能也存在着国家与投资者之间、公司非流通股东与流通股东之间、机构投资者委托人与管理人之间的代理问题。在代理机制主导下,当羊群行为作为一种策略独立于学习行为而存在时,策略性的羊群行为有助于增强非流通股东的谈判势力,增加流通股东的心理压力;有助于降低对价比例,或者有助于使既定的对价决策更容易获得流通股东的接受。这样,在股权分置改革的过程中,羊群行为与学习行为一样是具有“效益”的。与前人相关研究相比,本文提供了全新的视角与方法。首先,在大前提上,以虚拟博弈为研究框架乃是本文不同于前人对股权分置改革研究的根基,它决定了学习行为和羊群行为假说。其次,在研究主旨上,本文尝试以股权分置改革为样本,研究公司财务领域内的环境、动机与行动之间的相互影响;而前人有关公司决策的研究大多局限于分析公司特征对决策的影响,忽略环境影响,视决策过程为静态、单调的过程。第三,在研究范式上,本文乃是行为主义的视角和一般化条件:信息非完备、市场非完备以及人的有限理性假设;而前人有关学习行为的研究受样本条件的限制,一般都需要假设资本市场完备。第四,在研究思路和方法上,本文不拘泥于现有文献,而是立足于行为主义,借鉴社会心理学、行为博弈等相关学科的研究成果,采用推理、建立理论模型、分析数据与实证检验等手段,力争保证研究的科学、客观与严谨。第五,在研究内容上,本文首次进行了股权分置改革的学习行为和羊群行为的理论与实证研究;首次进行了羊群行为影响因素的理论和实证研究;首先进行了学习行为与羊群行为的比较研究。本文拓宽了公司财务学有关学习行为的理论与实证研究,也为全面评价中国上市公司股权分置的改革补充了学术证据。  
    [1] 严格地说,本文所研究的样本数据并非时间序列数据,因为每家公司只参加了一次对价决策博弈。本文“时间序列”一词仅约定俗成地表示股权分置改革在时间上的批次秩序。 [英文摘要]:     Abstract   The Chinese list companies’ non-tradable equity reform is a milestone-event in the capital market history. Within the non-tradable equity reform process, there appears three special characteristics: First is the local authorities and the national capital committees are actively performing the influence; Secondly is that non-tradable equity reform executed set by set - from few companies to a big range of companies; Thirdly the non tradable equity shareholders make the compensation plan first, and then the tradable equity shareholders vote on the plan instead of two parties jointly making the compensation plan at the same time. In the reform, the local authorities and the national capital committee’s active involvement may result in a deviation of the shareholder-right from Pareto optimization. A set-by-set process has made the early announced compensation plans become models for the later reforming companies. The model pressure possibly will influence the non-tradable equity shareholders’ negotiation power and the tradable equity shareholders’ negotiation power respectively. On average the non-tradable equity shareholders make the compensation plan one month before the shareholders’ meeting. In this period, more companies’ compensation plans come out with reference to the early companies, these new compensation plans will then re-influence the early companies tradable equity shareholders voting psychologically. So many facts prove that the specific of non-tradable equity reform process determine that the influences on the non-tradable equity shareholders and tradable equity shareholders are out of ratio and the negotiation power is gradually more favorable to non-tradable equity shareholders. So it looks like the process of the reform is a two-party game between the non-tradable equity shareholders and the tradable equity shareholders, but in fact, it is a decision-making process completely managed by the non-tradable equity shareholders. In the series of the decision-making, the tradable equity shareholders could not play their roles as they should, and the reform becomes a pseudo-player game where the interaction for the parties are among the tradable equity shareholders from different companies. So we would like to study the compensation decision-making behavior in the non-tradable equity reform with the theories of pseudo-player game including learning theory and herding theory. There are three noticeable characteristics during the non-tradable equity reform: the compensation ratio has an obvious clustering characteristic; the compensation ratio decreases gradually; and the 4-year bear market ended and the bull market begins. So can learning explain the compensation ratio’s decrease trend? Is the compensation ratio cluster come out of learning-convergence or herding? Why do the companies herd? Will the stock market influences the compensation-ratio decision? The market change is more favorable to which one, the non-tradable equity shareholders or the tradable equity shareholders? Is the compensation ratio’s clustering characteristics also playing a role here? Through analysis on the compensation decision-making process, we sets the pseudo-player game hypothesis for the non-tradable equity reform. Based on psychological, economic and financial theories, we establish the theory frame of learning and herding, and then provide proofs of “learning hypothesis” and “herding hypothesis” for the non-tradable equity reform, using the data of non-special treatment companies listed in Shanghai stock market and Shenzhen stock market as samples. We compare and analyze learning, including best-response learning and imitation learning, and herding and its internal and external influencing factors, such as the company’s characteristics, the stock price over-valued, the reform stage and the participants’ quantity. The empirical research provides evidences for learning and herding hypothesis. Theoretically learning means that in the observable environment, by means of analyzing individual’s experience, environmental structure or other people’s behavior or outcome, people improve their own decision information and then the decision utility. Learning roots in information asymmetry and human being’s limited rationality. While herding means in order to improve decision utility, such as making right decisions, acquiring selfish interest etc., decision makers give up private information, and imitate previous people or follow the group’s actions. Herding derives from information asymmetry, social norm or principal-agent problem. The correlation and difference between learning and herding are as: both of learning and the herding theory study group decision and behavior, and pay attention to the decision makers’ interactions and the environment’s influence on the decision makers. Herding always initially come from learning incentive, while learning catabolizes gradually as herding-behavior. Social learning improves the public information pool via group interaction to spread private information to all decision makers. On the contrary, herding deletes or weakens private information and finally prevents the public information poll from growing. Emerged in empirical outcome, learning may makes the group behavior converge or diverge. If convergence arises, and the learning chain is long enough, the convergency value is close to the truth. Herding must make the group behavior converge while it does not get close to truth. Under the informational mechanism, learning improves decision quality step by step, so the group utility increases gradually. While herding makes the decision quality worse, so the group utility decreases gradually (under the agent mechanism, individual utility might get improved by herding). By establishing a theoretic learning model of the non-tradable equity reform, we analyze out that in the serial, repetitive and complex process, there exits imitative learning and also optimum-response learning. The optimum-response learning includes not only reinforcement-based learning - through observation to reinforcement the early actor’s strategy attraction but also belief-based learning - through observation to change personal belief. The empirical study provides evidence for the theoretic analysis. in empirical study, we found that in the process of the non-tradable equity reform, there exists a symbiosis of optimum-response learning and imitative learning. The optimum-response learning explains the decreasing trend of compensation ratio. As for the imitative learning, in time series, it makes the reforming companies in neighboring set present stickiness in compensation ratios which brings partially the compensation ratio’s a cross-sectional clustering effect in the next set. However, the stickiness accumulation does not impact the compensation ratio’s decreasing trend, so cascades do not happen. Our empirical study also find that if the non-tradable equity ratio is smaller, the companies earning is better, the institutional investors’ shareholding is higher, the added commitment in the compensation plan is stronger, or the stock price is higher (relative to EPS), the non-tradable equity shareholders will pay less compensation. Obviously, when the stock market cake is expended bigger, it is more favorable to the non-tradable equity shareholders - they successfully made the compensation ratio lower so they get more benefit. Another interesting thing is that, being as the biggest tradable equity shareholders,  the institutional investors have not bargained a higher compensation ratio, on the contrast, they played the opposite role – the more their shareholdings, the less the compensation ratio. We also analyzed herding behavior in the non-tradable equity reform process with theoretical model and empirical research. By establishing theoretical model, we draw the following conclusions: the economic environment, industry competition, decision makers’ rationality and other factors will influence the decision quality; decision makers have herding incentive to get more information or psychological support. In the group of competition if the project asset value is overvalued by the stock market, herding will always happen. If the asset value is undervalued in the stock market, the decision makers will become more conservative, there may not herd. Decision makers’ attention to the project risks and the precise budget for the project will reduce the probability of herding. Through the empirical research on herding of the non-tradable equity reform, we found that: (1) There is no herding in time series in the non-tradable equity reform. But in the cross section, in addition to the compensation ratio cluster induced by learning, there also exits herding in the compensation decision-making. It shows that when deciding how much to compensate, the decision makers not only learn from prior companies’ decision method, but also imitate with each other in the same set. Besides, they also consider the company’s profitability, non tradable equity ratio, institutional shareholdings ratio and the additional commitment in the compensation and the like. The company size and the stock price-increase are not considerable; they may get implicit in learning or herding. (2) The political meddling in non-tradable equity reform restricts the decision makers’ choices space and independence, so there is nearly no way for them to be free from herding in the compensation decision process. It stands to reason that it also distorted the relationship between decision-making herding and the capital asset valuing. (3) Those companies with higher profitability provide lower compensation ratio and are more incline to adopt herding. It means that they are trying to reduce the compensation as much as they can, so they would like to follow the crowd to get support. Adopting herding in the compensation decision process can help to strengthen the tradable equity shareholders’ mental pressure and weaken their negotiation power. On the other hand, those companies with lower profitability are more conservative in the decision in view of their higher risk. The non-tradable share ratio does not significantly influence herding. This maybe because the average level of non-tradable share ratio is high enough; the equity reform will not threaten the big shareholder’s control on company. The more the public company’s debt, the less herding will be in the compensation decision. (4) Herding in the compensation ratio decision is negatively correlated to the numbers of companies in each set. It shows that when there is no competition and the group size is big enough, herding will be weakened. (5) As the reform going on, herding becomes stronger and stronger. Considering the phenomenon of “as the reform going on, the compensation ratio gets lower and lower”, herding presents its strategic effect: to adopt herding help to strengthen the tradable equity shareholders’ mental pressure and weaken their negotiation power, it will make the tradable equity shareholders more maneuverable to accept the lower compensation ratio.  As the non-tradable equity reform progresses, the pubic companies’ acknowledgement of herding effects spreads out. There exits informational asymmetry in the non-tradable equity reform. Besides, there also exits agent problems between national power and investors, major shareholders and small shareholders, principals and agents (especially in the governance of institutional investors). Under the agent mechanism, where herding acts as a strategy which can be independent of learning, the strategic herding behavior will help strengthen the non-tradable equity shareholders’ negotiation power and strengthen the tradable equity shareholders’ mental pressure, so reduce the compensation ratio or make it easier for the tradable equity shareholders to accept the compensation plan. So herding and learning are both creating “benefit” in the process of non-tradable equity reform. Comparing to the previous studies, this paper provide a brand new study view and method. First of all, using the pseudo-player game theory as the framework is what makes this paper different from other papers that study on non-tradable equity reform. It determines learning and the herding hypothesis. Secondly the theme of this paper is to use the non-tradable equity reform as samples to study the interaction of environment, incentive and actions in behavioral corporate finance. While previous studies on companies strategy decision is usually limited to an analysis of the companies characteristics’ effect on the decision-making, which neglect the environment and consider the decision process a static, monotonous process. Thirdly the research paradigm stands on behaviorism and generic conditions: incomplete information, incomplete market and limited rationality. But previous studies on learning usually suppose a perfect market being limited to their samples. Fourthly the analyzing method is not limited to the current literatures, but on the basis of behaviorism, with reference to social psychology, behavioral game theory. This paper uses inference, theoretical model, data analysis and empirical verification to ensure the study is scientific, objective and precise. Fifthly, it is the first time to do a theoretical and empirical study of learning and herding on the non-tradable equity reform. This paper is also the first one to do a theoretical and empirical study on the influence factors of herding, and the first one to give a comparative analysis of learning and herding. It extends the theoretical and empirical study about learning in corporate finance, and provides additional academic evidence to comprehensively evaluate Chinese non-tradable equity reform.      
关于我们 | 老师招聘 | 版权声明 | 联系我们 | 付款方式 | 返回顶部 | 

COPYRIGHT ©2001 - 2013 51LUNWEN.NET. ALL RIGHTS RESERVED.
【免责声明】:本网站所提供的信息资源如有侵权、违规,请及时告知
无忧论文网提供毕业论文指导 硕士论文指导服务