sort by 'Cross country industry year' model

[Ruba AB]

I am trying to modify a model I use to calculate earnings management, the model is perfect but for a cross-sectional sample for one country. I want to modify the model to work across various countries-Industries of each country - and year finding coefficients to each country-industry-year. The model is as below, I tried to add to sortby the country but I am not sure if this is enough, when I run the same model over one country alone (using a separate dataset includes only the specific country variables) the coefficients of the residual were not the same, still I am not sure if it should be the same:

gen z2_coef=.
gen z3_coef=.
gen z4_coef=.
gen cons_coef=.


gen R2=.
gen AjR2=.

gen fitted=.
gen resi =.

gen n1=.
gen n2=.

by SIC year, sort: gen normaln =_N


forval i=1/89 {

       forval y=2013/2021 { 
       
       count if SIC==`i' & year==`y'
       return list
       if `r(N)' >=3 {
       
         capture reg z1 z2 z3 z4 if SIC==`i' & year==`y'
         
           capture matrix a = e(b)
         
gen aaa if SIC== `i' & year==`y'= a[1,1]
gen bbb if SIC== `i' & year==`y'= a[1,2]
gen ccc if SIC== `i' & year==`y'= a[1,3]
gen ccc1 if SIC== `i' & year==`y'= a[1,4]

    replace z2_coef= aaa if SIC== `i' & year==`y'
    replace z3_coef= bbb if SIC== `i' & year==`y'
    replace z4_coef= ccc if SIC== `i' & year==`y'
    replace cons_coef= ccc1 if SIC== `i' & year==`y'
     drop aaa 
     drop bbb
     drop ccc
     drop ccc1
     
             
     
predict iii if SIC== `i' & year==`y'
predict jjj if SIC== `i' & year==`y', resid
   replace fitted= iii if SIC== `i' & year==`y'
   replace resi = jjj  if SIC== `i' & year==`y'
      drop iii
      drop jjj
          
          
          
ereturn list 
gen ggg if SIC== `i' & year==`y'= e(r2)
gen hhh if SIC== `i' & year==`y'= e(r2_a)
   replace R2= ggg if SIC== `i' & year==`y'
   replace AjR2= hhh if SIC== `i' & year==`y'
     drop ggg 
     drop hhh 
     
     
gen kkk if SIC== `i' & year==`y'= e(N)
gen mmm if SIC== `i' & year==`y'= _N
         replace n1= kkk if SIC== `i' & year==`y'
      replace n2= mmm if SIC== `i' & year==`y'
        drop kkk 
        drop mmm
          
                  }
             }
       }

:

[Ruba AB]

I have updated the following code, it works and gets me the residuals but still, I am not sure if it is correct or not.

gen z2_coef=.
gen z3_coef=.
gen z4_coef=.
gen cons_coef=.
gen R2=.
gen AjR2=.
gen fitted=.
gen resi =.
gen n1=.
gen n2=.

by country_id SIC year, sort: gen normaln =_N

forval i=1/89 {
    forval y=2013/2021 {
        count if SIC==`i' & year==`y' & country_id!=. 
        return list
        if `r(N)' >=3 {
            capture reg z1 z2 z3 z4 if SIC==`i' & year==`y' & country_id!=. 
            capture matrix a = e(b)
            gen aaa = a[1,1] if SIC== `i' & year==`y' & country_id!=.
            gen bbb = a[1,2] if SIC== `i' & year==`y' & country_id!=.
            gen ccc = a[1,3] if SIC== `i' & year==`y' & country_id!=.
            gen ccc1 = a[1,4] if SIC== `i' & year==`y' & country_id!=.
            replace z2_coef = aaa if SIC== `i' & year==`y' & country_id!=.
            replace z3_coef = bbb if SIC== `i' & year==`y' & country_id!=.
            replace z4_coef = ccc if SIC== `i' & year==`y' & country_id!=.
            replace cons_coef = ccc1 if SIC== `i' & year==`y' & country_id!=.
            drop aaa bbb ccc ccc1
            predict iii if SIC== `i' & year==`y' & country_id!=.
            predict jjj if SIC== `i' & year==`y' & country_id!=., resid
            replace fitted = iii if SIC== `i' & year==`y' & country_id!=.
            replace resi = jjj  if SIC== `i' & year==`y' & country_id!=.
            drop iii jjj
            ereturn list 
            gen ggg = e(r2) if SIC== `i' & year==`y' & country_id!=.
            gen hhh = e(r2_a) if SIC== `i' & year==`y' & country_id!=.
            replace R2 = ggg if SIC== `i' & year==`y' & country_id!=.
            replace AjR2 = hhh if SIC== `i' & year==`y' & country_id!=.
            drop ggg hhh
            gen kkk = e(N) if SIC== `i' & year==`y' & country_id!=.
            gen mmm = _N if SIC== `i' & year==`y' & country_id!=.
            replace n1 = kkk if SIC== `i' & year==`y' & country_id!=.
            replace n2 = mmm if SIC== `i' & year==`y' & country_id!=.
            drop kkk mmm
        }
    }
}