Event Study Tests : A Brief Survey

In this paper, I describe some of the main parametric and non-parametric tests used in event studies to assess the significance of abnormal returns or changes in variance of returns. The parametric tests here described are Brown and Warner's test (1980, 1985), with and without crude independence adjustment, Patell's (1976) standardised residual test and Boehmer, Musumeci and Poulsen's (1991) standardised cross-sectional test. The non-parametric tests are the generalised sign test (Sanger and McConnell, 1986 and Cowen and Sergeant, 1996), the Wilcoxson signed rank test and Corrado's (1989) rank test. In addition I outline briefly a bootstrapping procedure for the purpose of testing non-normal distributed abnormal returns. I also describe four tests to evaluate the significance of changes in variance. I present two parametric tests (an F test for the equality of two population variances and the Beavers's/May's U test) and two non-parametric tests (the squared rank test and another rank test proposed by Rohrbach and Chandra, 1989)