data df1;
   set df;
       by id;
       retain flag_final;
       if first.id then do;
          if flag eq 0 then flag_final=flag;
          else flag_final=.;
       end;
       else do;
          flag_final=flag_final;
       end;
       new_col=id;
   run;

df_temp=df[['id','flag']]
  df1= df.merge(df_temp.groupby('id').first().reset_index(), on='id').rename(
columns={'flag_x':'flag', 'flag_y':'flag_final'})