Dear community,
I'm struggling at the moment identifying which fixed effects I need to include in my two-way fixed effects (Diff-in-Diff) estimation. My data is FDI flows between country pairs per year. My event/treatment is the Brexit referendum, so I want to examine how FDI flows to Great Britain changed after the referendum compared to before the referendum relative to how FDI flows to all other ("untreated") countries changed after the referendum compared to before the referendum. To estimate such a Diff-in-Diff with panel data, I need to include time-, unit-FE as well as the treatment indicator variable (=1 for all observations where UK is FDI host country and year is >= 2016) (Wooldridge, 2021). I'm struggling to decide whether the unit FE need to be country-pair FE or host-country FE. My treatment varies at host country-& year-level but also at country-pair- & year-level. I would appreciate any hint!
Best
Noemi
I'm struggling at the moment identifying which fixed effects I need to include in my two-way fixed effects (Diff-in-Diff) estimation. My data is FDI flows between country pairs per year. My event/treatment is the Brexit referendum, so I want to examine how FDI flows to Great Britain changed after the referendum compared to before the referendum relative to how FDI flows to all other ("untreated") countries changed after the referendum compared to before the referendum. To estimate such a Diff-in-Diff with panel data, I need to include time-, unit-FE as well as the treatment indicator variable (=1 for all observations where UK is FDI host country and year is >= 2016) (Wooldridge, 2021). I'm struggling to decide whether the unit FE need to be country-pair FE or host-country FE. My treatment varies at host country-& year-level but also at country-pair- & year-level. I would appreciate any hint!
Best
Noemi
Comment