Collapsing panel dataset given an if condition

[Federica Vassalli]

Hello everybody,
i'm struggling with what it seems to me a very silly question.

I have a panel dataset in which Date is my t and ateco is my id variable.

* Example generated by -dataex-. To install: ssc install dataex
clear
input float Date long ateco double(PDy emp gva out mkp)
153 1                 . 1144.8 8394.2 90.4 106.9
154 1                 . 1121.7 8455.1 90.1 107.4
155 1 1.733790447524462 1103.1 7853.6 90.4 103.9
156 1 1.653791042927907 1079.2 8387.1 90.2 108.3
157 1 1.619407230835545 1084.5 8549.3   88 107.9
end
format %tq Date
label values ateco ateco
label def ateco 1 "A", modify

Ateco has 10 different values, as you can see here

  ATECO |      Freq.     Percent        Cum.
------------+-----------------------------------
          A |         94       10.00       10.00
          B |         94       10.00       20.00
          C |         94       10.00       30.00
         DE |         94       10.00       40.00
          F |         94       10.00       50.00
        GHI |         94       10.00       60.00
          J |         94       10.00       70.00
       KLMN |         94       10.00       80.00
        OPQ |         94       10.00       90.00
        RST |         94       10.00      100.00
------------+-----------------------------------
      Total |        940      100.00

I am interested in ateco C, GHI and F and I want to collapse data (by Date) for all the other ateco in one new ateco, i.e. W, so that I end up with a panel dataset where Date remains the t dimension and the id var ateco has 4 values, C - GHI - F - W.

I've tried with a dummy variables with no success. Would you mind to help me?

[Andrew Musau]

decode ateco, g(str_ateco)
replace str_ateco="W" if !inlist(str_ateco, "C", "GHI", "F")
collapse PDy emp gva out mkp, by(str_ateco Date)

[Federica Vassalli]

Thank you so much!! It works much better than creating dummies