kylebutts / did2s_stata

Two-Stage Difference-in-Differences following Gardner (2021)
30 stars 14 forks source link

Fix what happens when every unit is eventually treated #12

Open kylebutts opened 1 year ago

kylebutts commented 1 year ago

Currently, the first_stage regression will produce point estimates in the periods after everyone is treated (assuming those fixed effects are zero). This is not correct and should instead drop those years all together and use the last treated group as the "control group"