【作者】 杨华；

【导师】 陈迅；

【作者基本信息】 重庆大学 ， 数量经济学， 2011， 博士

【副题名】基于股权分置改革前后面板门限模型的比较分析

【摘要】 资本资产定价模型(CAPM)和有效市场假说是经典金融理论的基石,主要探讨完全竞争市场和理性经济人假设条件下资本市场的行为。对国内外研究分析资本市场中收益和风险之间的关系产生了深远的影响。CAPM认为影响股票期望报酬率的唯一因素是系统风险,并且揭示出两者之间呈正向的线性关系。然而,近三十年来国内外越来越多的研究证实存在许多无法用传统的资产定价理论解释或者与有效市场假说相违背的现象,尤其是股票市场异象的产生,说明我们不能够在经典的资产定价理论框架下探讨股票价格的内在运行规律,除了系统风险之外,还可能有其他的因素可以解释股票异常收益的现象,例如:规模效应、盈余价格比效应、账面市值比效应等。许多学者如Fama﹠French等试图运用不同的研究方法、选取不同国家、不同时期的样本数据、不同的研究视角,在有效市场假说的前提下给出种种解释,然而现有的理论对这些问题并不能完全解释清楚。从另外一个角度看,它们也为我们深入探究资本市场中股票价格的形成与变化规律提供了一个新的视角或途径,因而这些股票市场异象一经发现,就成了理论界与实务界关注的焦点,因为要解释这些异常现象需要发展新的理论和方法,这在一定程度上也推动金融学科的发展。正是基于这样一种考虑,近三十年来,国内外有许多学者选取不同国家、不同时间段、不同的样本数据对股票市场异象从不同角度进行分析研究;但大多数研究所使用的方法一个共同的特点是研究者根据他主观想法按某一特征值(比如账面市值比,公司规模等)对样本分组构建不同的投资组合。但很少有学者对于我国A股市场运用面板门限模型(Hansen,1999)分析检验各项股票市场异象。运用面板门限模型对比分析检验股权分置改革前后沪深A股市场规模效应和账面市值比效应的存在性、形成原因、以及股权分置改革对沪深股市有效性的影响。一方面可以帮助我们加强理解沪深A股市场自身结构特征、市场交易机制、市场参与者行为以及交易价格形成机制等影响证券市场行为的微观层面问题;另一方面通过比较两市的实证结果找到两个市场运行的特点、存在的差异,为监管层更好地认识两市运行规律提供实证证据,由此考察股票异象随时间变化的演变规律。有鉴于此,本文运用面板门限模型,选取1996年12月31日前已上市的深圳A股、上海A股部分上市公司为研究对象,以2005年4月29日为实质性股权分置改革的分界点,在考虑异质预期影响的情况下分析比较股权分置改革前后我国股票市场的异象,检验规模效应、账面市值比效应是否存在、简要分析异象产生的原因及这两种异象之间是否存在交叉作用;并通过纵向比较股改前后两市实证检验结果,来检验股权分置改革对沪深A股市场效率的影响。现将论文的主要研究结论总结如下:①在股改以前的1997年1月—2005年4月时间段内,研究结果表明:整体而言,沪市A股规模效应显著存在、账面市值比效应不明显。深市A股规模效应和账面市值比效应都显著存在。而股改后的2008年1月—2010年10月时段内。研究结果表明:沪市A股规模效应依然存在着显著,而账面市值比效应不存在。深市A股规模效应显著存在,而账面市值比效应不再存在。②股改前,沪市A股公司规模存在双重门限效应,公司规模与股票收益率在三个区间依次存在正、负、负相关关系,在相关关系为正的区间,统计上不显著,样本公司大多数落在后两个区间,占样本总数的90.01%,统计上显著但解释程度不同;账面市值比存在单一门限效应,账面市值比与股票收益率在两个区间依次存在负、正相关关系,在负相关关系区间显著,但正相关关系区间统计上不显著。深市A股,公司规模不存在门限效应;账面市值比存在单一门限效应,账面市值比与股票收益率在两个区间依次存在负、正相关关系,在相关关系为负的区间,统计上不显著;而在相关关系为正的区间,存在显著的正相关关系,该区间样本占样本总数55.39%。股改后,沪市A股公司规模存在单一门限效应,公司规模与股票收益率在这两个区间都为显著的负相关关系,但解释程度不同;账面市值比不存在门限效应。深市A股公司规模存在单一门限效应,公司规模与股票收益率在两个区间依次为正相关、负相关关系,但解释程度和显著性不同,前者统计上不显著,后者统计上显著,后者区间样本占样本总数84.34%;账面市值比不存在门限效应。因此,我们分析股权分置改革前后沪深市A股市场规模效应、账面市值比效应时,要区别异象因子在不同区间的不同变化趋势。在不同的门限区间内公司规模、账面市值比对股票收益率的解释程度不同,前者主要是由于不同规模的公司被庄家操控股价、并购重组的难易程度不一样,活跃程度也不一样;后者主要由于不同账面市值比的公司基本面不一样,稳定性、股价波动程度不一样。③公司规模的影响。检验结果表明,股改前后深市A股公司规模对股票收益率的解释力和显著性都要比沪市A股整体要强;股改前后沪深股市各自检验结果纵向比较表明股改后各自公司规模对股票收益率的解释程度也增强了。④账面市值比的影响。股改前后沪市A股四个样本区间,只有股改前深市A股账面市值比对股票收益率有较强的解释力,与股票收益率存在显著的正相关关系。其他情况下,账面市值比效应不明显。⑤异质预期的影响。检验结果表明,股改前后沪深A股四个样本区间异质预期对股票收益率的影响都存在,对股票收益有明显的预测作用,其代理变量换手率与股票收益率都为显著的正相关关系,即表明预期差异程度越大对股票收益收益率影响也越大。股改前后深市异质预期对股票收益率的解释程度都要比沪市A股要强。纵向比较结果表明股改后沪深A股预期差异程度对股票收益的解释力与股改前相比都减弱了。⑥交易金额。股改前检验结果表明沪市A股成交金额对股票收益率产生显著的负向影响,表明交易金额小的股票获得的股票收益率更高。与沪市A股相比,深市A股中该指标对股票收益率的影响也为负相关关系,但是深市A股,解释程度要强。股改后的检验结果表明:沪深两市成交金额对股票收益率都产生显著的负向影响,即对股票收益有预测作用,深市该指标对股票收益率的解释程度要比沪市仍然要强。⑦公司规模、账面市值比的交叉作用。股改前后沪深A股四个样本区间检验结果都表明公司规模、账面市值比两异象因子不存在交叉作用。⑧纵向比较结果表明:整体上,股权分置改革促进了沪深A股市场效率的改进。⑨我国股票市场不太成熟,理性投资缺乏,投机性强,庄家操纵严重股价,小规模公司是并购重组的热点,也是投资者追捧的热点。整体而言,股权分置改革并未完全改变沪深股市高投机性的特性,股改前后,深市A股市场投机性、不稳定性都要强于沪市A股市场。本文主要创新点如下:①不同于已有文献大多构建横截面数据模型或者是时间序列数据模型,针对这一问题,本文则构建了面板数据模型,这样能够识别一些前面两种模型所不能识别的因素提高了模型参数估计精度。在此基础上,为了克服人为分组而导致分析结果的偏差,本文尝试运用面板门限模型来改善这项缺失,根据样本资料本身的信息找出门限分区点,区分出非线性关系的区间,这样不仅可验证不同门限区间内股票异象是否存在,而且可探讨公司规模、账面市值比对股票收益率解释程度的差异。检验结果表明,在不同的区间(存在门限或不存在门限),表现出不同的规律;并且在不同的门限区间,由于不同的公司规模(账面市值比)公司被操纵、并购重组的难易程度不一样(基本面不一样,稳定性不一样),所以导致门限变量(公司规模或账面市值比)对股票收益率解释程度不同,与大多已有研究结论单一规律存在差异。改进了目前国内外分析检验股票市场异象所使用的方法,大多是研究者把研究样本按特征值由低到高排序构建不同的投资组合进行分析的做法。②在已有的研究中,我国学者大多独自构建相应的股票投资组合作为其实证研究对象,样本选取时间段短,大多选用2005年以前的沪市A股数据,并且不同的学者对于市值规模的界定不一致,难以覆盖市场整体情况,造成其重复验证性以及研究的可持续性都比较差,针对这一问题,本文同时选取沪深A股股改前后四个时间段的样本进行分析检验。通过纵向比较股改前后沪深A股检验结果来分析股权分置改革对股票市场效率的影响。结果表明股改前后,深市投机性氛围都要比沪市确实要强烈,股权分置改革在一定程度上促进了沪深A股市场效率的改进。③考虑异质预期的影响,把换手率作为异质预期的代理变量。改进已有大多数研究较少考虑异质预期的影响,此外还拓展了仅单独检验一种股票异象的情况,更综合的分析了规模效应、账面市值比效应两种异象之间的关系,尝试检验两种异象因子的交叉作用。在借鉴参考Fama&French三因子模型的基础上选取公司规模、账面市值比、成交金额、换手率、交叉作用因子构建新的模型分析股改前后沪深A股异象。检验结果表明股改前后沪深A股异质预期的影响都存在,预期差异程度与股票收益率正相关,股改前后预期差异程度对股票收益率的解释程度深市都要比沪市强烈,股改减弱了异质预期对两市的影响;规模、账面市值比不存在交叉作用。更多还原

【Abstract】 Capital asset pricing model (CAPM) and efficient market hypothesis, which are the foundation of classical finance theory, mainly discuss the behavior of capital markets based on perfectly competitive market and the ’rational man’ hypothesis.They had a profound impact domestic to the relationship between return and risk in capital market around theory circle and practice circle at home and abroad. CAPM account the only factor that affects the stock’s expected return is the systematic risk, exists positive linear relationship between them. However, during the last three decades, more and more research confirmed that there are still so many inexplicabilities away from the traditional CAPM or goes against with the efficient market hypothesis when maintaining stock market equilibrium. In particular, the anomalies of stock market illustrated that we can not entirely study the internal operating rules of stock price under classic framework of CAPM. Besides systematic risk, there may are still some other factors could explain the phenomenon of stock abnormal returns. Such as: size effect, earnings-to-price effect, book-to-market effect, overreaction, underreaction, weather effect, etc. Many scholars, such as Fama & French, etc. try to use different study methods, selecting different countries, different periods of sample data, different study perspectives to give the explanations under efficient market hypothesis. However, the existing theory couldn’t give a reasonable explanation for these problems.From another point of view, they also provided us a new perspective or approach to deeply study the stock price formation and variation rules in the capital market. Therefore, the anomalies of stock market once are discovered, they would became the focus of theory circle and practice circle. To explain these anomalies need to develop new theories and methods, which also contributed to the development of financial discipline to some extent. Based on this kind of consideration, during the last three decades, many scholars at home and abroad selected different countries, different time intervals, different samples data to study on the anomalies of stock market from different angles. But most of all the methods used for the study had a common feature that researchers would build different portfolios based on a particular characteristic value (such as book- to- market, size, earnings-to-price, etc.) by their subjective idea. However, especially for A share market, few scholars chose panel threshold regression model (Hansen, 1999) to analysis the anomalies of stock market. After nearly two decades of booming development, A-share market has accumulated plenty of operating data to provide precious data foundation for our study on the size effect and book- to- market effect of two markets. In addition, a large number of foreign relevant theory provides a good theoretical guidance for our present study. Using new nonlinear measurement methods: (panel threshold) to comparative analysis before and after the split share structure reform of A-share market size effects and book- to- market effect existence, causes, and the impact of A-share market by split share structure reformation. On the one hand, it can help us strengthen the A-share markets structural characteristics, market trading mechanisms, behavior of market participant and market pricing mechanism, etc. which influence the behaviors in stock market at the micro-level, the differences between Shanghai stock market and Shenzhen stock market. And then, by comparing the empirical results of two market, the operation features of two markets were found and the empirical basis of two markets’operation law was provided to regulators for studying the regularity and robustness of stock market anomalies changing over time.For this reason, by utilizing the non-linear measurement methods - panel threshold model (Hansen ,1999), this thesis selects a part of listed companies up to December 31, 1996 on A share market as research objectives, taking April 29, 2005 as the cut-off point of substantive share-split reform, considering the impact of Heterogeneous expectation to analysis and compare the anomalies of stock market before and after the split share structure reform, to test the existence of the size effect, book- to- market effect , a brief analysis of the causes of anomalies and interrelationship of two anomalies, and then, tests the affection of split share structure reform to A share market effectiveness.The main conclusions of this paper are as follows:Firstly, Between January 1997 to April 2005, prior to the split share structure reform, research shows that: on the whole Shanghai A Stock market has significant size effect, and no obvious book- to- market effect, while Shenzhen A Stock market has significant size effect and also obvious book- to- market effect. However, after the completion of the split share structure reform between January 2008 to October 2010, research shows that: on the whole, Shanghai A stock market still has significant size effect, but no book- to- market effect, while Shenzhen A stock market has significant size effect, but no book- to- market effect .Secondly, It is tested that prior to the split share structure reform, In Shanghai A stock market company size has double threshold. It is positively, negatively, and negatively correlated in the three sample areas respectively with stock return.It is not statistically significant in the positive area.Most of the sample companies are in the latter two areas, statistically significant but to different degree of explanation,accounting for 90.01% of the total samples. Book- to- market has single threshold effect, has negative, positive correlation with stock return in two interval, but not statistically significant in the positive area. In Shenzhen A stock market, company size has no threshold effect, and book- to- market has single threshold effect. book-to-market has negative, positive correlation with stock return in the two interval. In the negative area , no significant statistics; but significant in the positive area, accounting for 55.39% of the total samples.After the split share structure reform , In Shanghai A stock market, company size has single threshold, is significantly negative correlated with stock returns in the two interval, but degree of explanation is different. Book-to- market has no threshold effect. In Shenzhen A stock market, company size has single threshold, is positively, negatively correlated with stock return in the two interval , accounting for 55.39% of the total samples. but not statistically significant in the positive area, Book-to- market has no threshold effect. Therefore, when we do the analysis for size effect and book-to-market effect in Shanghai and Shenzhen stock market before or after the reform, we need to differentiate the different trend of in different area.In different threshold areas . company size, book-to-market on stock returns are explained by the degree is different, the former is mainly due to the different scale of the company stock price manipulation by the declarer, merger and reorganization of the difficult easy degree is different the active degree is not the same; the latter mainly due to different book to market value than the company fundamentals are not the same, stability volatility of stock price, not the same degree.Thirdly, the influence of company size . Test results show: before or after the split share structure reform, company size in Shenzhen A stock market has more significant explanation on the stock return than those companies in Shanghai A stock market.Before or after the split share structure reform,Longitudinal comparison tell us the role of company size on stock return in both market has been enhanced.Fourthly, the influence of book-to- market . Before or after the split share structure reform, in the four samples area in Shanghai and Shenzhen A stock market,only book-to- market in Shenzhen A stock market has obviously positive relationship with stock return.In other cases, therefore book-to- market effect is not obvious. Fifthly, the influence of heterogeneous expectation . Test results show that, before or after the split share structure reform, in the four samples area in both Shanghai and Shenzhen A stock market,heterogeneous expectation has positive impact on the stock returns.It plays a significantly predictive role on it.Besides,the proxy variable exchange rate and stock return are positively cross impacted, that means, the higher the exchange rate of company stock is, the greater the stock return is. Before or after the split share structure reform heterogeneous expectation has great effect on stock return in Shenzhen A stock market than in both Shanghai A stock market , The longitudinal comparison in Chapter 7 shows that after the split share structure reform, in both Shanghai and Shenzhen A stock market, heterogeneous expectation has less effect on stock return.Sixthly, trading volume. Test results show that before the split share structure reform, the trading volume in Shanghai A stock market plays a significantly negative role on stock return, which means the less the amount is,the higher the return is. Compared with Shanghai A stock market, in Shenzhen A stock market,it has the same negative relationship with stock return, but more obvious than that in Shanghai A stock market.After the split share structure reform test results show that the trading volume in Shanghai A stock market play a significantly negative role on stock return. However,Shenzhen A stock has greatly impact than Shanghai A Stock.Seventhly, The interactions of company size and book-to-market.Before or after the split share structure reform,the four sample area in Shanghai and Shenzhen stock market tell us that company size is not related with book-to-market.Eighthly, Longitudinal results showed that: Overall, the split share structure reform to some extent contributed to the Shanghai and Shenzhen A-share market efficiency improvementsNinthly, China’s stock market is not mature, irrational, speculative, making a serious price manipulation, small-scale mergers and acquisitions the company is hot, but also investors hit. Overall, the split share structure reform has not completely change the Shanghai and Shenzhen stock markets highly speculative characteristics, before and after the split share structure reform, Shenzhen A-share market speculation, instability are stronger than the Shanghai A-share marketThe main innovative points are as follows:①Different from the existing literature mostly construction cross section data model or time series data model, To solve this problem, this paper constructs a panel data model, this can identify some in front of two kinds of models can identify factors to improve the accuracy of the estimation of model parameters. on the basis of,In order to overcome the subjective packet causes the analysis result the deviation, this paper attempts to use panel threshold model to improve this deficiency, according to the sample data information threshold partition points out, to distinguish the asymmetric relationship between area, so that it not only can show different threshold area size effect, book-to-market effect exists, and can explore different threshold within the range of company size, book to market on stock returns to explain the extent. The test results showed that, in different areas (the existence of threshold or absence of threshold ), showed a different pattern; and in different threshold areas, due to the different company size (book-to-market) was manipulated, merger and reorganization of the degree of difficulty is not the same ( fundamentals are not the same, stability, so.) led to the threshold variable (company size or book-to-market) on stock returns explain to different degrees, and most previous research conclusions there is a significant difference in a single rule. This is change that Analysis of stock anomalies of the method used, mostly based on their subjective idea of the study sample according to the feature values from low to high ranking construct different investment portfolio analysis, because the artificial distinction sort will lead to different conclusions.②In previous studies, most Chinese scholars to construct the corresponding stock portfolio as the case study object, sample selection time short, chosen mostly in 2005 before the Shanghai A stock data, and different scholars for market scale definition of inconsistency, to cover the overall market, caused by the repeated verification and research sustainability is poorer, To solve this problem ,this paper selects the Shanghai and Shenzhen A shares before and after the split share structure reform, four time of samples for analysis. Through comparing the test results Shanghai &Shenzhen A shares to the analysis of the split share structure reform on the stock market efficiency. The results show that before and after the split share structure reform, Shenzhen speculative atmosphere than Shanghai to strongly, Overall , the split share structure reform contributed to the Shanghai & Shenzhen A-share market efficiency improvements.③This paper considers the heterogeneity effect, and perceives the turnover as the proxy variable of heterogeneous expectations. Improvement for most studies seldom consider the heterogeneity effect.Currently,most research only examine such type stock vision solely. However, on the basis of this, this paper analyzes the relationship between size effect and book-to-market effect , and try to test two anomalies factor interactions. Based on the threefactor model of Fama&French, this paper selects company size, book to market value ratio, volume, turnover , interactions factor to establish the new model for analysis test before and after the Shanghai and Shenzhen A shares anomalies. The test results showed that before and after the Shanghai and Shenzhen A shares heterogeneity effects are present, the expected difference in degree and stock returns are related, before and after the expected difference in degree on stock returns to explain the extent of Shenzhen than Shanghai strongly, the weakened heterogeneous expectations on two city influences; size, book-to-market does not exist interactions.更多还原