Abstract
Event study is a powerful tool for analyzing the dynamic effects of policy and other shocks in microeconomics. However, there is little understanding of how to apply this method when individuals or locations experience multiple events in close succession. We explore methods of estimating a multiple event study with Monte Carlo simulations. Allowing multiple event-time dummies to be turned on at once generally produces unbiased estimates, while ignoring subsequent events or duplicating observations to have one observation per individual-event-time create trends in the outcome variable before and after an event that can be misleading to the researcher. We present empirical applications which show that the choice of method can make important differences in practice.