Regression on countries subsamples with country Fixed effects

[marie nour]

Dear statlisters,

I have a question on how stata includes the fixed effects in a regression with a subsample.
I have multiple country pairs and compute the fixed effects like this:

    * Fixed effects
    egen imp = group(importer)       // importer id
    egen exp    = group(exporter)    // exporter id
    rowsort imp exp, gen(simp sexp)  // sort by row so that order does not matter
    egen pair = group(simp sexp)     // pair id;  order does not matter
    drop simp sexp

Then, I run a regression with the inclusion of importer and exporter Fixed effects for a subsample (only for country pairs that involves South countries, which corresponds to bilateral ==3):

 reghdfe ln_imports ln_dist comlang_off contig if  bilateral==3, absorb(imp exp) vce(cluster pair)

My questions are: in the regression, are all FE included, or only FE on the subsample (i.e. South-South country pairs)? What are the implications in terms of estimates?
Could I use instead interaction terms so I include all the sample?

Thank you.
Regards