事件模拟与非正常收益模型的检验力——基于中国A股市场的经验检验

本文分别以1990年12月至2003年12月两市所有A股公司为抽样总体,检验了均值调整模型、市场调整模型和市场模型为基础的多种检验方法的检验力。研究发现,无论事件研究中各公司事件是否相近或重叠,都应采用市场模型为基础的非参数秩检验法。而累积非正常收益的检验也应以市场模型为计算基础。若样本公司事件日相近或重叠,统计量的设置要考虑累积非正常收益截面数据的相关性。均值调整模型在本文所定义的各种检验方法中,均无明显优势。经敏感性测试,本文结论不变。

Event Simulation and the Test Power of Abnormal Return Model

We evaluate the performance of alternative test statistics in the event studies which include Chinese daily security returns.Whether there is clustering in event dates,we find that the nonparametric rank statistic performs the best overall in our samples;we recommend its use with market model abnormal returns to test event day abnormal returns.For tests over the long interval,the test statistic should consider the cross-sectional dependence in the security abnormal returns.Sensitivity tests do not change our conclusions.