MEASURING SECURITY PRICE PERFORMANCE USING DAILY NASDAQ RETURNS

We evaluate the performance of alternative test statistics in event studies which include NASDAQ daily security returns. We document varying degrees of test statistic misspecification in NASDAQ samples. In particular, we find that the commonly used standardized test statistic is misspecified in most settings. Although less pervasive, misspecification is also evident in the portfolio test statistic estimated using the time series of portfolio mean abnormal returns. The nonparametric rank statistic [introduced in Corrado (1989)] performs the best overall in NASDAQ samples; we recommend its use with market model abnormal returns based on an equal-weighted NASDAQ market index.