Ways to test for reverse causality and omitted variables in stata with difference-in-differences

[Guest]

Hello statalist

I am conducting a difference-in-differences analysis in stata. the setup is a simple 2x2 DiD setup so i estimate the model like so: REG Y D##P, cluster(city). D is my treatment indicator (binary) P is my pre/post indicator (binary)

are there any ways to test for reverse causality and ommited variables in such a setup in stata? i was thinking about including some pre treatment controlvariables to probe the assumptions of parrelel trends but otherwise i am at a loss.

kind regards

[Maxence Morlet]

These two articles should help:

https://arxiv.org/abs/2201.01194

https://papers.ssrn.com/sol3/papers....act_id=3906345