Hello guys in Statalist!
I have some problems with estimation on trade creation and diversion effects. I am using the command ppml_panel_sg with trend options. That means the model includes it jt and ij fixed effects as well as a exporter-importer-specific time trend.
Now I would like to estimate the Trade Creation and Trade Diversion effect, and I set two "single member" dummies to indicate whether source country (i) and destination country (j) is a member of the trade agreement (BRI) respectively. In other words, BRI_i takes value of 1 if i is a member and j is not, zero otherwise. BRI_j takes value of 1 if i is not a member and j is a member, zero otherwise. When I estimate these two dummies, one of them will be omitted due to collinearity.
I find some papers mention that to address the collinearity issue we need to drop the exporter-year(it) and importer-year(jt) FE. My question is that how to estimate these two "single member" dummies excluding it and jt FE. Is it possible to use ppml_panel_sg command? Or I need to use command ppmlhdfe? If so, it seems there is no option to add this exporter-importer-specific time trend in ppmlhdfe command, then how to set this variable?
Thanks everyone in advance!
I have some problems with estimation on trade creation and diversion effects. I am using the command ppml_panel_sg with trend options. That means the model includes it jt and ij fixed effects as well as a exporter-importer-specific time trend.
Now I would like to estimate the Trade Creation and Trade Diversion effect, and I set two "single member" dummies to indicate whether source country (i) and destination country (j) is a member of the trade agreement (BRI) respectively. In other words, BRI_i takes value of 1 if i is a member and j is not, zero otherwise. BRI_j takes value of 1 if i is not a member and j is a member, zero otherwise. When I estimate these two dummies, one of them will be omitted due to collinearity.
I find some papers mention that to address the collinearity issue we need to drop the exporter-year(it) and importer-year(jt) FE. My question is that how to estimate these two "single member" dummies excluding it and jt FE. Is it possible to use ppml_panel_sg command? Or I need to use command ppmlhdfe? If so, it seems there is no option to add this exporter-importer-specific time trend in ppmlhdfe command, then how to set this variable?
Thanks everyone in advance!
Comment