Using joint regression as an alternative to suest in panel data

[Jose Ferreras]

I am trying to test if the coefficients of (c. approp_c and c.approp_c#c.approp_c) in the following two models are statistically different.

xtreg breadth_colln c.approp_c c.approp_c#c.approp_c c.L.breadth_colln c.int_rd c.grupo c.size c.market_size if gr, vce(cluster ident) fe
xtreg breadth_colln c.approp_c c.approp_c#c.approp_c c.L.breadth_colln c.int_rd c.grupo c.size c.market_size if !gr, vce(cluster ident) fe

For estimating the model I am using xtreg, but suest cannot be applyied after xtreg. Therefore I am trying to apply Andrew Musau suggestion to obtain joint estimation: https://www.statalist.org/forums/for...ta?view=thread

xtreg breadth_colln i.gr#(c.approp_c c.approp_c#c.approp_c c.L.breadth_colln c.int_rd c.grupo c.size c.market_size), vce (cluster ident) fe

In the example introduced by Andrew, the coefficients in the individual regressions are the same as the one obtained in the joint regression. However, I don’t understand why in my case the coefficient that I obtain in the individual and in the joint regression are differents. Don't they should be the same?

[FernandoRios]

The reason is that your fixed effects are not being "interacted" with GR.
That being said, you could check this https://friosavila.github.io/chatgpt...fe_07_14_2023/
for some ideas how to apply SUEST to this scenario
HTH