A Re-Evaluation of Event-Study Methodology

This paper provides evidence through observations and simulations that Cumulative Abnormal Return (CAR) can result in misleading inferences about market efficiency and post-event behavior. A set of 96 companies known to have multiple events and a simulation of returns with multiple post-event events are used to test three invalidating hypotheses on event-study methodology. We find that the use of artificial portfolios in event studies biases CAR downward, the increased volatility around post-event events and on the event day lowers the significance of abnormal returns, and the time series of CAR of the individual securities is larger in magnitude and higher in significance than that of the cross-sectional CAR of the portfolio.