Event Window Parameters Estimation with Application to M&A Deals

The paper addresses a long-standing problem of event window choice arbitrariness in event studies. We develop a framework for rigorous event window parameters estimation based on three groups of methods and a validation procedure for the resulting estimates. This is achieved by incorporating advanced statistical machinery into the established event study structure. We also ensure that our framework does not interfere with existing statistical significance testing methods. Implementation of the framework is illustrated on a set of US M&A deals from late 2009 to 2017. We find that our estimated event windows are substantially shorter than fixed event windows commonly used in literature. Evidence suggests that researchers tend to underestimate the speed of market reaction to news and introduce errors into their results by choosing unnecessarily long windows. Our framework can be readily utilized by other researchers seeking to reduce arbitrariness of event window choice in their studies.