panel regression with country group year fixed effects

[Seng Yee]

Hi everybody,

For my research, I need to do multiple FE with panel data.

the dependent variable is group-level, Independent variables is country level.

So I tried

reg y x i.group i.country i.year

it response with not sorted, can't proceed.

Only in this circumstances is possible to proceed:

egen ID = group(country group)
xtset ID year
reg i.group i.year i.country

I am not sure did I formulated the right code.

Thanks

[Carlo Lazzaro]

Seng:
your -fe- regression can be easier with -xtreg,fe-.
By the way, you do not have to -xtset- your dataset if you go -regress-.
If you're interested in n-way -fe- you should consider the community-contributed module -reghdfe-.

[George Ford]

reghdfe y x , absorb(country year)

you should think about collapsing the data to the country level if all your X's are country level.