Difference in differences
A Causal inference method.
When there is a Natural experiment (e.g., an implementation of a policy only in certain counties), we can compare the outcomes between those with enacted policy and those without. If we assume that there is a Parallel trend across all places, then we can compute the difference (between treatment and control) of the differences (within treatment or within control) and attribute it to the policy’s causal impact.
It is challenging to apply the DID method when many things are happening simultaneously. It is difficult to disentangle them. (e.g., see Zimmerman2021association, Goodman Bacon2021using)
So you’re trying to choose an event-study estimator…