Improving the Efficiency of Event Studies under a Random Walk Hypothesis
by Bryan Engelhardt

An average cumulative abnormal return statistic (CAR) used in event studies to test whether an event impacts a firm's stock price will equal zero given a random walk hypothesis. As a result, many event studies such as Mackinlay(1997) categorize the events relative to market expectations prior to calculating a CAR. We show this split in the data is statistically inefficient relative to alternative statistics and the market expectation data introduces measurement error. Furthermore, we demonstrate the impact of the inefficiency with Monte Carlo simulations and explore its impacts on well accepted results in the event study literature. Note, the alternative approach can affect 1000+ event study results.

Click here for a detailed report.

Stage of idea: Requisition

Unmet requisite(s): Statistician

Academic discipline(s): Statistics, Finance, Event Studies

Created on Oct 14, 2017 and last updated on Nov 07, 2017.

Suggested citation: Engelhardt, Bryan, 2017. "Improving the Efficiency of Event Studies under a Random Walk Hypothesis.",

Post a comment | Post a private comment to author(s)