Hi
I am studying the effect of choosing a policy on the stores of some companies; as a result, I have more than 2000 treatment groups. I also have over 10000 control stores (nearby and same industry store). Because of many treatment and control stores and as I do not have many covariates to match the stores, I think that the only option is to use classic DID. However, the parallel assumption does not hold. I was wondering whether there is any analysis that helps me to dodge, like any robustness check that verifies that parallel assumption is not a threat here.
I am studying the effect of choosing a policy on the stores of some companies; as a result, I have more than 2000 treatment groups. I also have over 10000 control stores (nearby and same industry store). Because of many treatment and control stores and as I do not have many covariates to match the stores, I think that the only option is to use classic DID. However, the parallel assumption does not hold. I was wondering whether there is any analysis that helps me to dodge, like any robustness check that verifies that parallel assumption is not a threat here.
