【作者】 蒋清中；

【导师】 吕长江；

【作者基本信息】 吉林大学 ， 企业管理， 2012， 博士

【副题名】基于盈余预测去偏及股价泡沫过滤

【摘要】 当投资者预计投资报酬率超过其要求的必要报酬率（即权益资本成本）时，投资者才会选择该投资目标。这意味着若不能正确估计权益资本成本，就很可能会作出错误的选择。隐含权益资本成本指的是使资产价值与未来预期收益现值相等的折现率，其估计是否准确，有赖于估计模型的准确性、盈余预测的可信度和股票内在价值的合理性。然而，如果分析师的盈余预测数据存在着系统性偏差，或者以股票价格来代替其价值，都会影响隐含资本成本估计结果的正确性。同时，不同的隐含权益资本成本估计模型计算的结果存在着较大差异，难以进行取舍，这就增加了投资者决策的难度，也加大了企业以权益资本成本作为投资决策依据的不准确性。股票价格中包含了投资者对企业前景的预期和对股票市场波动的预期，既然权益资本成本是长期投资的依据，就应当将股价随市场波动而发生的短期波动剔除。本研究以文献回顾的结果为基础建立了权益资本成本的估计框架，此框架包括三个步骤：一是盈余预测去偏，即事前去除分析师盈余预测存在的系统性偏差得到市场盈余预期；二是个股泡沫过滤，即滤去股票价格中的泡沫得到股票的内在价值的期望值；三是隐含权益资本成本估计，即以事前去偏的盈余预测和滤去泡沫的股票价格为基础，选择最优方法计算隐含权益资本成本。本研究的原始数据来自于CSMAR数据库、中经数据库、凤凰网，将原始数据导入到SQL Server数据库后，用SQL语言对数据进行了初步加工，并使用SPSS16.0进行了回归分析，使用EViews 6进行了时间序列分析。对1998年至2010年上证综合指数进行测算后，发现2005年5至12月与2008年8至10月等期间内，上证综合指数低于市场基本价值，处于负泡沫区域。2007年10月上证综合指数远高于市场基本价值。对2003年至2010年沪深A股1134个上市公司进行了计算，与分析师盈余预测原始数据相比较，0期、1期、2期盈余预测事前去除偏差后，其偏度下降了9.33%、8.79%、14.47%。以去偏后盈余预测和滤泡沫后股票价格为基础，DIV1、DIV2、OJ、PFE、RIV1、RIV2等六个估计模型计算出的隐含权益资本成本平均值分别为12.97%、16.42%、14.59%、6.42%、6.47%、10.77%，将隐含权益资本成本与未来实现报酬相联系，PFE模型估计的权益资本成本与未来一个季度、两个季度等一直到未来三年的累计实现报酬均一致地呈现出显著正相关关系（P<0.001），检验结果显示盈余预测去偏且股价滤去泡沫后的PFE模型具有最强的预测能力。与此相反，在仅对盈余预测去偏，或仅对股价泡沫过滤，或直接采用原始盈余预测与股票价格等三种不同条件下，PFE模型估计值与未来实现报酬则不具有长期一致的正相关关系，这验证了本研究提出的隐含权益资本成本估计框架的有效性。本研究的发现为实务界估计权益资本成本提供了一个合理的框架，即必须考虑到分析师盈余预测的系统性偏差及股票价格的泡沫度等因素，并选择PFE模型作为首选估计模型。同时，本研究拓展了套利定价模型（ATP），原模型将影响股票收益率的宏观经济因素和企业特有风险因素放在同一个层面上进行研究，本研究则建立了从宏观经济因素影响股票市场到股票市场影响个股收益的分层研究模型。本研究的创新之处有二：一是提出了个股内在价值的测度方法，即运用协整检验及误差修正模型建立宏观经济变量与股票市场综合指数的长期均衡关系模型，以指数对均衡的偏离计算出股票市场泡沫，进而根据个股泡沫与市场泡沫之间的关系求出个股泡沫，最后以个股价格减去个股泡沫倒挤出个股价值。二是建立了隐含权益资本成本估计的完整框架，即盈余预测去偏、个股泡沫过滤、估计模型优选等三个步骤和方法。更多还原

【Abstract】 Only when investors forecast that the rate of return on investment is more thanrequired rate of return, i.e. cost of equity capital, can investors choose the investmentobjective.This means if they can not estimate the cost of equity capital accurately,investors might make a false option.Implied cost of capital refers to the discount rate which makes the assets valueequal the present value of the expected earning in the future. The accuracy of theevaluation depends on the accuracy of the estimated model, the reliability of theearnings forecast and the rationality of the stock intrinsic value. However, if there issystematic bias for analyst’s earning forecast or stock value is replaced by stockprice, the result’s correctness of the implied estimate of cost of capital will beinfluenced. Meanwhile, there is a big difference on the calculation result of differentestimate models of implied cost of equity capital. It’s hard to for investors to make adecision.Stock price includes investors’expectation to enterprise’s prospect and theexpectation to the fluctuation of the stock market. Since cost of equity capital is thebasis of long-term investment, stock’s short-term fluctuation which occurs whenstock price fluctuates along with the market should be removed. The study builds aframework of estimating the cost of equity capital based on the outcome of literaturereview: firstly, de-biased analyst forecast, i.e., remove the systematic deviation ofanalyst earning forecast in advance to get the market earning forecast; secondly,stock bubble filter, i.e., filter the bubbles in stock price and get the expected value ofstock intrinsic value; thirdly, the estimation of implied cost of equity capital, i.e., optimizing a method to calculate the implied cost of equity capital based onde-biased earning forecast in advance and the stock price filting bubbles.The original data had been retained from CSMAR database, Chinese Economicdatabase, ifeng.com website and put into SQL Server database to be processed there.The general statistical analysis including regression analysis has been completed bySPSS16.0. As for the processing of time series data ,EViews 6 is applied to completeit.This research measured and calculated Shanghai Composite Index (SCI) during1998 to 2010. During the period of from May to December, 2005 and from August toOctober, 2008 and so on, SCI was lower than the basic market value and was in anegative bubble range. SCI was far higher than the basic market value in October,2007. This research calculated 1134 listed companies for A stock in Shanghai andShenzhen between 2003 and 2010 and compared the result with the original date ofanalyst’s earning forecast. The skewness decreased by 9.33%, 8.79% and 14.47%respectively for phase 0, phase1, phase 2 when the the deviation of the earningforecast was removed in advance; The mean of the implied cost of equity capital is12.97%,16.42%,14.59%, 6.42%, 6.47% and 10.77% respectively calculated by the sixestimation models: DIV1, DIV2, OJ, PFE, RIV1 and RIV2 based on de-biasedearning forecast and the stock price after the bubbles have been removed. Connectingthe implied cost of equity capital with the future realized returns, cost of equitycapital and the accumulative reward realized in next quarter, next two quarters and tillnext three years conformably have a significant positive correlation(P<0.001)estimated by PFE model. The findings show that PFE model with both debiasedearning forcast and filtered stock price has the strongest forecast capacity. In contrast,PFE model and the future realized returns don’t have a long-term consistent positivecorrelation when neither of the earging forcast nor stock price having been treated,which has verified the validity of the framework. In practice, this study provides a reasonable framework of estimating the cost ofequity capital. It is very important to remove the bias of analyst earnings forecastsand the bubbles of stock price, and the PFE model should be the preferred one.Also, this study extends the arbitrage pricing model (ATP), which investigateseffects of macroeconomic factors and firm-specific risk factors on the stock returns toat the same level. this study establish a hierarchical model to analyze the effect ofmacroeconomic factors on the stock market and the one of the market on each stock.There are two innovations in this study. The first one is the measure of theintrinsic value of a stock. Cointegration tests and error correction model are used toexplore the long-run equilibrium relationship between the macroeconomic variablesand SSCI, and then the stock market bubble was calculated based on the deviationfrom the equilibrium.Stock bubble is derived from stock price to get its intrinsic value.The secone one is the creation of a complete three-step framework of the implied costof equity capital estimation including the earnings forecast bias removing, the stockbubbles filtering, and the selecion of the optimized estimation model.更多还原