管理学研究方法管理学研究方法Event Study西南财经大学西南财经大学 会计学院会计学院唐雪松唐雪松Email: tangxs@1授课:XXXReferences•Brown and Warner (1980), JFE, 205-258.•Brown and Warner (1985), JFE, 3-31.•Bowman(1983), JBFA, 561-584.•Boehmer, et al (1991), JFE, 253-272.•Patell (1976), JAR, 246-276.•Kothari (2001), JAE, 105-231.•Campbell, J. et al. 1997. The Econometrics of Financial Markets. Princeton University Press. 上海财经大学出版社(中文版)•S.P. Kothari, Jerold B. Warner. 2006. Econometrics of Event Studies. in the Handbook of Corporate Finance: Empirical Corporate Finance, by B. Espen Eckbo.2授课:XXX一、引子3授课:XXX回顾:市场效率•不同信息对价格具有不同的影响程度,因而存在三种不同程度的市场效率:–弱式有效–半强式有效–强式有效•事件研究可以应用于分析市场是否半强式有效,即证券价格是否会立即反映公开信息对公司价值的影响。
4授课:XXX5授课:XXX事件研究法的作用–1. Information content (Ball and Brown)–2. Market efficiency (FFJR)–3. Model evaluation–4. Metric explanation6授课:XXX一些例子7授课:XXX8授课:XXX9授课:XXX10授课:XXX事件研究法的应用范围事件研究法的应用范围(1)公司财务领域:如兼并收购的财富效应以及公司融资决策的价格效应等2)经济法律领域:如被美国高等法院接受作为在内幕交易案中重大性原则的专业性判断标准,并作为审理证券欺诈性时确定受偿水平的参照方法3)宏观经济调控:如用于测定规制监管环境的变化对公司价值的影响等4)……11授课:XXX文献:举例文献:举例1. Halpem (1982)1. Halpem (1982)::利用事件研究的方法对公司并购进行了相关的研究,验证了有关公司并购的一些财务理论 2. 2. Manuel, Manuel, Pilotte(1996)Pilotte(1996):利用事件研究的方法分析了前次股票分割对收益率的影响,并据此预测本次股票分割对收益率的影响,结果证明可以利用前次事件对收益率的影响来合理地预测类似事件再次发生所造成的影响。
此时,预测收益率的变动不仅要考虑股票分割事件本身,而且要考虑前次股票分割后收益变化情况这一因素12授课:XXX3.El-Gazzar (1998)3.El-Gazzar (1998)::研究表明:机构投资者持股比例越高,跟进公司的金融分析师越多,收益公告的信息效应越弱4. Steiner4. Steiner,,Volker(2001)Volker(2001)::研究了欧洲债券价格的变化与 Watch listings 和Standard & Poor 关于信用评级的公告之间的联系,发现:(1)当出现信用评级下降时,债券价格将有显著的变化;而出现信用评级上升时,并没有很显著的宣告效应2)特别地,当信用等级降为投机级(speculative grade)时,价格变化尤其显著13授课:XXXPublications•事件研究已有很长一段历史,也许从事这一研究首次出版著作的人是多利(Dolley,1933)•20世纪60年代末,由鲍尔和布朗(Ball and Brown,1968)以及法玛、费雪、简森和罗尔(Fama,Fisher,Jensen and Roll,1969)所进行的富有创造性的研究中引入了此研究方法。
14授课:XXXPublications•According to the census of event studies published in 8 leading journals (1969-2005): Journal of Business (JB), Journal of Finance (JF), Journal of Financial Economics (JFE), Review of Financial Studies (RFS), Accounting Review (AR), Journal of Accounting & Economics (JAE), Journal of Accounting Research (JAR), and Review of Accounting Studies (RAS), the number of published event studies exceeds 800, and the literature continues to grow. Since many academic and practitioner-oriented journals are excluded, these figures provide a lower bound on the size of the literature.15授课:XXX二、概念与原理•A study of the relationship between a particular event or class of events and the response of a variable of interest. –For example, a researcher may be interested in the relationship between company stock repurchases and stock returns. The researcher measures the returns in the period prior to and after the day a stock repurchase is announced for a sample of securities. Although the dates of the stock repurchase announcements differ across time, the pre-event and post-event periods are uniform in relation to the event. The researcher tests to see if the average returns in the pre-event and post-event periods are significantly different from each other. 16授课:XXX基本原理基本原理 假设市场理性,则有关事件的影响将会立即反映在证券价格之中。
于是,运用相对来说较短期的所观察到的证券价格就可以测定某一事件的经济影响 17授课:XXX三、事件研究法的步骤(1) 事件定义(Event definition)(2) 取样标准(Selection criteria)(3) 界定正常和非正常收益率 (4) 参数估计(Estimation procedure)(5) 检验(Testing procedure) (6) 实证结果(Empirical results)(7) 解释和结论(Interpretation and conclusions) 18授课:XXX(1)事件定义(Event definition)•确定所要研究的事件•明确事件的研究期间—事件窗 (event window).–进行事件研究的第一步就是对重要事件进行定义,并确定该事件涉及到的公司的证券价格进行考察的时期,这一时期称为“事件窗”(或事件期)•例如,如果一个人在查看包括每日数据的收益公告的信息内容时,事件就是收益公告,事件窗则可能是公告中的某一天实际上,事件窗经常被延长到两天,即公告的当天和公告的第二天19授课:XXX(2) 取样标准(Selection criteria)–可获得性–避免选样偏差20授课:XXX(3)界定正常和非正常收益 •正常收益是指假设不发生该事件条件下的预期收益。
•非正常收益即事件期间内该证券事前或事后实际收益与同期正常收益之差21授课:XXX(4)参数估计(Estimation procedure)•界定估计窗 (estimation window) (estimation window) (event window) 0•应用正常收益模型进行参数估计22授课:XXX(5)检验(Testing procedure) –得到正常收益的参数估计后,计算非正常收益需要重点注意原假设的设定以及各公司非正常收益的汇总6) 实证结果(Empirical results)–表达与诊断7) 解释和结论(Interpretation and conclusions)–可以补充一些附加分析23授课:XXX四、正常收益估计模型与应用四、正常收益估计模型与应用•统计模型•经济模型24授课:XXX常见的正常收益估计模型:统计模型常见的正常收益估计模型:统计模型 (1).市场模型(market model)•即假定市场收益与证券收益之间存在稳定的线性关系•优点:剔除了收益中与市场收益波动相关 的部分,从而降低了估计误差。
•但是,该模型有效性依赖于可决系数R方, R方越大,则可以剔除越多的变异,收效越大25授课:XXX•市场模型中26授课:XXX(2).market-adjusted return model•以市场收益率作为个股的正常收益率,即在market model 中,令 该模型应用于无法事前估计模型参数的情况,不得已才使用例如,IPO 中27授课:XXX(3)常量均值模型(Constant Mean Return Model)即假定某一证券的平均收益不随时间的改变而改变Brown and Warner (1980, 1985) 发现该模型与其他更复杂的模型所计算的结果是非常相似的,原因可能在于非正常报酬的方差并不容易通过复杂的模型降低28授课:XXXmarket model VS Constant Mean Return Model (Mackinlay,1997)29授课:XXXMean-adjusted return model• 正常收益是股票前几期的平均收益率 30授课:XXX(4)因素模型(factor model)•one-factor model(e.g. market model)•multifactor model (Fama-French 3 factor: beta, size, BM; Carhart 4 factor: beta, size, BM, and Momentum).31授课:XXXmultifactor model的问题•the gains from employing multifactor models for event studies are limited. •the marginal explanatory power of additional factors is small, and hence, there is little reduction in the variance of the abnormal return. •The variance reduction will typically be greatest in cases where the sample firms have a common characteristic, for example they are all members of one industry or they are all firms concentrated in one market capitalization group.32授课:XXX常见的正常收益估计模型:经济模型常见的正常收益估计模型:经济模型•CAPM–The use of the Capital Asset Pricing Model is common in event studies of the 1970s. The validity of the restrictions imposed by the CAPM on the market model is questionable. This has introduced the possibility that the results of the studies may be sensitive to the specific CAPM restrictions. Because this potential for sensitivity can be avoided at little cost by using the market model, the use of the CAPM has almost ceased.•the restrictions of CAPM–投资者都是在期望收益率和方差的基础上选择投资组合。
–投资者具有完全相同的预期且均投资组合理论来选择证券组合–在资本市场上没有摩擦33授课:XXX•Arbitrage Pricing Theory (APT).–A general finding is that with the APT the most important factor behaves like a market factor and additional factors add relatively little explanatory power. –The main potential gain from using a model based on the arbitrage pricing theory is to eliminate the biases introduced by using the CAPM. However, because the statistically motivated models also eliminate these biases, for event studies such models dominate.34授课:XXXExample(以市场模型为例)(以市场模型为例)(1). 求正常收益 从估计窗中运用OLS得到 35授课:XXX(2) .把 代入事件窗中求非正常收益AR 36授课:XXX(3)计算累计非正常收益率CAR与标准化非正常收益率SCAR 37授课:XXX(4)计算平均累计非正常收益率 38授课:XXX如:如:39授课:XXX40授课:XXX(5)统计量 41授课:XXX非正常收益的应用与检验•参数检验。
•横截面分析(Cross-Sectional Models):横截面分析属于对非正常收益更细致的研究,即通过回归分析研究各个公司的具体特征变量(如规模、盈利状况等)与非正常收益的关系例)•非参数检验–符号检验(The sign test (Cowan,1992))–秩检验(Corrado,1989)42授课:XXX例:Asquith 和Mullins(1986)把股票增发公告的非正常收益对增发规模(占公司总价值的比率)和公告前11个月的累积非正常收益进行回归,发现增发规模越大,则负的非正常收益越大;前11个月的累积非正常收益越大,则负的非正常收益越小43授课:XXX五、势分析(Analysis of Power)•An important consideration when setting up an event study is the ability to detect the presence of a non-zero abnormal return.•因此,有必要研究给定水平的非正常收益下,拒绝原假设的可能性有多大,即需要评价检验的势(Power of test)44授课:XXX显著性水平为a的势:45授课:XXX46授课:XXX47授课:XXX48授课:XXX抽样区间对势的影响49授课:XXX六、事件研究失效的可能原因1.证券市场可能是无效率的,非半强式有效市场。
2.公司的公告信息与公司价值是无关的3.事件信息已经被完全预期4.内幕交易存在(insider trading),信息只会影响普通投资者而不会影响内部人士5.难于精确识别事件日期 50授课:XXX七、总结•事件研究法的概念与原理•事件研究法的步骤•正常收益的估计•势分析•事件研究失败的原因51授课:XXX52授课:XXXThank you!53。