Could the rumor clarification pacify the stock market：Evidence from China

Our initial sample consists of all the Chinese A-share listed firm samples spans from January 1, 2014 to December 31. The 5-min frequency price data and the trade and quote data are collected from the CSMAR high-frequency database; the former data are used to calculate the RV. The clarifications are collected from the CNRDS database, we retains only the rumor clarified in pessimistic announcement as these rumors may represent malicious attacks and more likely to attract the managers’ keen attention, the clarification variable Clari,t takes the value of 1 if observing a clarification during the (t-15, t-1) trading days for firm i in day t, and 0 otherwise. We use the fraud punishments as the negative measurement of the companies’ reputation (Fraud), Fraudi,t is a dummy variable get the value 1 if there is fraud punishments happened during the past 6 years, which are collected from CSMAR database. We delete the firms without rumor clarification, the sample-firms going public and the firms with trading days less than two years, and the samples with missing variables. Finally we get 515137 daily samples comprising 599 firms, with the maximum length of 1219 trading days.本研究的初始样本涵盖2014年1月1日至2014年12月31日期间的全部中国A股上市公司样本。其中，5分钟频率价格数据与交易报价数据取自国泰安高频数据库（CSMAR），前者用于计算已实现波动率（Realized Volatility，RV）。澄清公告数据来自中国国家研究数据服务平台数据库（CNRDS）：本文仅保留在负面公告中予以澄清的谣言，此类谣言往往属于恶意攻击，更易引发上市公司管理层的高度关注。澄清变量Clari,t的取值规则为：若在第t日对应公司i的(t-15, t-1)交易日区间内观测到相关澄清公告，则Clari,t取1，否则取0。本文采用欺诈处罚作为企业声誉的反向测度指标（Fraud）：Fraudi,t为虚拟变量，若公司i在过去6年内发生过欺诈处罚事件，则Fraudi,t取1，该数据取自国泰安数据库（CSMAR）。本文剔除了无对应谣言澄清公告的公司、新上市样本、交易时长不足2年的公司以及存在缺失变量的观测样本，最终得到包含599家公司的515137条日度样本，样本最长观测周期为1219个交易日。