Daily Stock Returns, Non-Normality and Hypothesis Testing

Daily stock returns typically have non-normal and asymmetric distributions, potentially leading to problems with hypothesis testing based on reported probability statistics from regression analysis (Fama 1976; Brooks 2002). While daily stock return data for many years is readily available, recent studies indicate that the non-normality problems may persist even in large samples (Peiro 2002; Brooks 2002; Bai et al 2002; Corrado and Zivney 1992). Given the frequent use of daily return data with standard estimation methods such as least squares, the practical importance of nonnormality on traditional hypothesis testing is of some significance.