Analyses of the Distribution of Security Market Model Prediction Errors for Daily Returns Data
分析日收益率数据中市场模型预测误差（异常收益）的实际分布，发现其偏离正态性，并探讨非同步交易对普通最小二乘估计的影响，对使用事件研究法的学者有参考价值。
Many studies in accounting use the (one-factor) market model to examine the effects of firm-specific events on the prices of their securities. Early studies used monthly returns data, but more recently the use of daily data has become popular partly in order to take advantage of more powerful statistical tests available with these data. However, two problems are commonly encountered with the use of daily data: nonsynchronous trading (see Scholes and Williams [1977] and Dimson [1979]) and significant departures of the data from normality (see Fama [1965; 1976], Praetz [1972], Clark [1973], Blattberg and Gonedes [1974], and Marais [1984]). Scholes and Williams have shown that nonsynchronous trading leads to biased and inconsistent estimates of the market model parameters using ordinary least squares (OLS). These deficiencies will obviously affect the distribution of prediction errors (or residuals) and their statistical tests, which are the focus of information content studies. Moreover, nonnormality of returns results in misstatements of the significance levels in statistical tests which typically assume normality. The main purpose of this paper is to report the results of a detailed analysis of daily stock returns in which I compared the actual empirical distributions of standardized prediction errors (abnormal returns) with