Announcement

Collapse
No announcement yet.
X
Collapse
  • Page of 1
  • Filter
  • Time
  • Show
Clear All
new posts
  • #1

    Find changes within firms' structure in longitudinal data (ownership type & industry)

    Dear all,

    Greetings!

    I am struggling to find a way to count unique firms which could change their ownership type and industry throughout the time period in my dataset (2012-2018). I have created a unique id for each firm, but I am stuck at how I can identify whether there were some changes throughout the years.

    Any advice will be greatly appreciated!

    (okfs=ownership type; ind=industry)

    Code:
    clear
input float(id year) byte okfs float ind
1 2014 34 60
1 2015 34 51
1 2016 34 46
1 2017 34 46
1 2018 34 46
2 2012 23 70
2 2013 23 70
2 2018 23 68
3 2016 24 41
3 2017 24 41
3 2018 24 41
4 2017 24 41
5 2016 34 45
5 2017 34 45
6 2016 34 68
6 2017 34 46
7 2017 24 56
7 2018 24 56
8 2017 24 41
9 2016 24 47
9 2017 24 47
10 2016 34 46
10 2017 34 46
10 2018 34 46
11 2016 24 47
12 2013 21 85
13 2016 24 41
14 2016 24 47
14 2017 24 25
14 2018 24 25
15 2016 24 1
15 2017 24 1
16 2016 23 1
16 2017 27 1
16 2018 27 1
17 2016 23 1
17 2017 27 1
17 2018 27 1
18 2016 24 47
18 2017 24 47
18 2018 24 47
19 2016 24 41
20 2016 34 45
20 2017 34 45
20 2018 34 45
21 2016 24 81
21 2017 24 81
21 2018 24 81
22 2016 34 70
23 2016 24 81
24 2016 24 46
25 2016 24 46
25 2017 24 46
26 2016 23 46
27 2016 24 46
27 2017 24 46
28 2014 21 85
28 2015 21 85
28 2016 21 88
28 2017 21 88
28 2018 21 88
29 2016 24 41
29 2017 24 41
30 2016 24 56
30 2017 24 56
30 2018 24 56
31 2016 24 46
31 2017 24 46
31 2018 24 46
32 2016 34 3
32 2017 34 3
33 2016 34 71
33 2017 34 71
33 2018 34 71
34 2016 23 1
34 2017 27 1
34 2018 27 1
35 2016 34 79
35 2017 34 79
35 2018 34 79
36 2016 24 8
37 2016 24 46
38 2015 18 91
38 2017 18 94
38 2018 18 94
39 2016 24 45
39 2017 24 45
39 2018 24 45
40 2016 34 28
40 2017 34 28
40 2018 34 28
41 2016 24 41
42 2017 23 46
43 2016 24 22
43 2017 24 22
43 2018 24 46
44 2016 34 74
44 2017 34 71
45 2016 34 41
45 2018 34 41
end
    Last edited by Iuliia Svetetskaia; 13 Mar 2022, 07:16.
    Tags: None
  • #2
    Thanks for the data example.

    See the FAQ https://www.stata.com/support/faqs/d...ions-in-group/

    There are at least two interpretations of change, (1) whether last and first values differ for each firm and (2) whether there has been any change within the period, which could mean a firm changing one way or another but ending at the same value as at first. The second is perhaps unlikely here, but here is code regardless.

    Code:
    clear
input float(id year) byte okfs float ind
1 2014 34 60
1 2015 34 51
1 2016 34 46
1 2017 34 46
1 2018 34 46
2 2012 23 70
2 2013 23 70
2 2018 23 68
3 2016 24 41
3 2017 24 41
3 2018 24 41
4 2017 24 41
5 2016 34 45
5 2017 34 45
6 2016 34 68
6 2017 34 46
7 2017 24 56
7 2018 24 56
8 2017 24 41
9 2016 24 47
9 2017 24 47
10 2016 34 46
10 2017 34 46
10 2018 34 46
11 2016 24 47
12 2013 21 85
13 2016 24 41
14 2016 24 47
14 2017 24 25
14 2018 24 25
15 2016 24 1
15 2017 24 1
16 2016 23 1
16 2017 27 1
16 2018 27 1
17 2016 23 1
17 2017 27 1
17 2018 27 1
18 2016 24 47
18 2017 24 47
18 2018 24 47
19 2016 24 41
20 2016 34 45
20 2017 34 45
20 2018 34 45
21 2016 24 81
21 2017 24 81
21 2018 24 81
22 2016 34 70
23 2016 24 81
24 2016 24 46
25 2016 24 46
25 2017 24 46
26 2016 23 46
27 2016 24 46
27 2017 24 46
28 2014 21 85
28 2015 21 85
28 2016 21 88
28 2017 21 88
28 2018 21 88
29 2016 24 41
29 2017 24 41
30 2016 24 56
30 2017 24 56
30 2018 24 56
31 2016 24 46
31 2017 24 46
31 2018 24 46
32 2016 34 3
32 2017 34 3
33 2016 34 71
33 2017 34 71
33 2018 34 71
34 2016 23 1
34 2017 27 1
34 2018 27 1
35 2016 34 79
35 2017 34 79
35 2018 34 79
36 2016 24 8
37 2016 24 46
38 2015 18 91
38 2017 18 94
38 2018 18 94
39 2016 24 45
39 2017 24 45
39 2018 24 45
40 2016 34 28
40 2017 34 28
40 2018 34 28
41 2016 24 41
42 2017 23 46
43 2016 24 22
43 2017 24 22
43 2018 24 46
44 2016 34 74
44 2017 34 71
45 2016 34 41
45 2018 34 41
end

bysort id (year) : gen diff_o = okfs[1] != okfs[_N]

by id : gen diff_i = ind[1] != ind[_N] 

bysort id (okfs) : gen diff_o_2 = okfs[1] != okfs[_N] 

bysort id (ind) : gen diff_i_2 = ind[1] != ind[_N] 

tabdisp id, c(diff*)


----------------------------------------------------------
       id |     diff_o      diff_i    diff_o_2    diff_i_2
----------+-----------------------------------------------
        1 |          0           1           0           1
        2 |          0           1           0           1
        3 |          0           0           0           0
        4 |          0           0           0           0
        5 |          0           0           0           0
        6 |          0           1           0           1
        7 |          0           0           0           0
        8 |          0           0           0           0
        9 |          0           0           0           0
       10 |          0           0           0           0
       11 |          0           0           0           0
       12 |          0           0           0           0
       13 |          0           0           0           0
       14 |          0           1           0           1
       15 |          0           0           0           0
       16 |          1           0           1           0
       17 |          1           0           1           0
       18 |          0           0           0           0
       19 |          0           0           0           0
       20 |          0           0           0           0
       21 |          0           0           0           0
       22 |          0           0           0           0
       23 |          0           0           0           0
       24 |          0           0           0           0
       25 |          0           0           0           0
       26 |          0           0           0           0
       27 |          0           0           0           0
       28 |          0           1           0           1
       29 |          0           0           0           0
       30 |          0           0           0           0
       31 |          0           0           0           0
       32 |          0           0           0           0
       33 |          0           0           0           0
       34 |          1           0           1           0
       35 |          0           0           0           0
       36 |          0           0           0           0
       37 |          0           0           0           0
       38 |          0           1           0           1
       39 |          0           0           0           0
       40 |          0           0           0           0
       41 |          0           0           0           0
       42 |          0           0           0           0
       43 |          0           1           0           1
       44 |          0           1           0           1
       45 |          0           0           0           0
----------------------------------------------------------

assert diff_o == diff_o_2 

assert diff_i == diff_i_2
    In the data example, the answer is the same for each original variable.
    • 2 likes

    Comment

    • #3
      Thank you so much, Nick!

      May I ask another question?

      Can I make this more conditional? So, for example, can I make it so the difference variable will be equal to 1 only if there was a change of ownership type (okfs) from "okfs==31 | okfs==32 | okfs==33 | okfs==34" to "okfs==21 | okfs==22 | okfs==23 | okfs==24 | okfs==27"?

      Thank you in advance!


      Code:
      clear
input float(id year) byte okfs float ind
 1 2014 34 59
 1 2015 34 51
 1 2016 34 46
 1 2017 34 46
 1 2018 34 46
 2 2012 23 69
 2 2013 23 69
 2 2018 23 68
 3 2016 24 41
 3 2017 24 41
 3 2018 24 41
 4 2017 24 41
 5 2016 34 45
 5 2017 34 45
 6 2016 34 68
 6 2017 34 46
 7 2017 24 55
 7 2018 24 55
 8 2017 24 41
 9 2016 24 47
 9 2017 24 47
10 2016 34 46
10 2017 34 46
10 2018 34 46
11 2016 24 47
12 2013 21 85
13 2016 24 41
14 2016 24 47
14 2017 24 25
14 2018 24 25
15 2016 24  1
15 2017 24  1
16 2016 23  1
16 2017 27  1
16 2018 27  1
17 2016 23  1
17 2017 27  1
17 2018 27  1
18 2016 24 47
18 2017 24 47
18 2018 24 47
19 2016 24 41
20 2016 34 45
20 2017 34 45
20 2018 34 45
21 2016 24 80
21 2017 24 80
21 2018 24 80
22 2016 34 69
23 2016 24 80
24 2016 24 46
25 2016 24 46
25 2017 24 46
26 2016 23 46
27 2016 24 46
27 2017 24 46
28 2014 21 85
28 2015 21 85
28 2016 21 87
28 2017 21 87
28 2018 21 87
29 2016 24 41
29 2017 24 41
30 2016 24 55
30 2017 24 55
30 2018 24 55
31 2016 24 46
31 2017 24 46
31 2018 24 46
32 2016 34  3
32 2017 34  3
33 2016 34 71
33 2017 34 71
33 2018 34 71
34 2016 23  1
34 2017 27  1
34 2018 27  1
35 2016 34 79
35 2017 34 79
35 2018 34 79
36 2016 24  5
37 2016 24 46
38 2015 18 90
38 2017 18 94
38 2018 18 94
39 2016 24 45
39 2017 24 45
39 2018 24 45
40 2016 34 28
40 2017 34 28
40 2018 34 28
41 2016 24 41
42 2017 23 46
43 2016 24 22
43 2017 24 22
43 2018 24 46
44 2016 34 74
44 2017 34 71
45 2016 34 41
45 2018 34 41
46 2016 24 41
46 2017 24 41
46 2018 24 41
47 2016 24 46
47 2017 24 46
47 2018 24 46
48 2016 34 46
48 2017 34 46
48 2018 34 46
49 2017 23  1
49 2018 23  1
50 2016 24 45
50 2017 24 45
50 2018 24 45
51 2016 24 47
51 2017 24 47
51 2018 24 47
52 2018 24 33
53 2016 24 96
54 2016 24  1
54 2017 24  1
54 2018 24  1
55 2016 23 46
56 2016 24 62
57 2016 24 41
58 2016 24 52
58 2017 24 52
58 2018 24 52
59 2016 24 47
60 2016 24 47
60 2017 24 47
60 2018 24 47
61 2016 24 80
62 2016 24  1
62 2017 24  1
62 2018 24  1
63 2016 24 55
64 2016 24 25
64 2017 24 25
64 2018 24 25
65 2016 23 68
65 2017 23 68
65 2018 23 68
66 2017 34 37
67 2016 24 46
67 2017 24 46
67 2018 24 46
68 2016 24 46
69 2016 34 46
69 2017 34 46
69 2018 34 46
70 2016 24  1
71 2016 34 62
71 2017 34 62
72 2016 24 55
73 2016 34 41
73 2017 34 41
73 2018 34 41
74 2017 24 41
75 2016 34 94
75 2017 34 94
75 2018 34 94
76 2016 34 46
77 2017 24 45
77 2018 24 45
78 2016 24 80
79 2016 23 46
79 2017 23 46
79 2018 23 46
80 2016 34 45
81 2017 24 41
81 2018 24 41
82 2016 24 45
82 2017 24 45
82 2018 24 45
83 2017 34 49
84 2014 24 45
85 2015 23 52
86 2018 24 80
87 2016 24 64
87 2018 24 64
88 2014 23 52
88 2015 18 52
88 2016 24 47
88 2017 24 47
89 2014 24 51
89 2015 24 51
90 2014 34 51
90 2015 34 51
91 2015 24 45
91 2016 24 45
91 2017 24 41
91 2018 24 45
92 2016 24 52
92 2017 24 52
92 2018 24 52
93 2015 24 51
93 2017 24 46
93 2018 24 46
94 2018 24  1
end

      Comment

      • #4
        Surely. I would construct a new variable first.

        Code:
        . gen which = cond(inrange(okfs, 31, 34), 1, cond(inlist(okfs, 21, 22, 23, 24, 27), 2, 3))

. tab okfs which

           |              which
      okfs |         1          2          3 |     Total
-----------+---------------------------------+----------
        18 |         0          0          3 |         3 
        21 |         0          6          0 |         6 
        23 |         0          8          0 |         8 
        24 |         0         46          0 |        46 
        27 |         0          6          0 |         6 
        34 |        31          0          0 |        31 
-----------+---------------------------------+----------
     Total |        31         66          3 |       100
        • 1 like

        Comment

        • #5
          Originally posted by Nick Cox View Post
          Surely. I would construct a new variable first.

          Code:
          . gen which = cond(inrange(okfs, 31, 34), 1, cond(inlist(okfs, 21, 22, 23, 24, 27), 2, 3))

. tab okfs which

| which
okfs | 1 2 3 | Total
-----------+---------------------------------+----------
18 | 0 0 3 | 3
21 | 0 6 0 | 6
23 | 0 8 0 | 8
24 | 0 46 0 | 46
27 | 0 6 0 | 6
34 | 31 0 0 | 31
-----------+---------------------------------+----------
Total | 31 66 3 | 100

          Thank you again, Nick!

          The thing is I have tried out these codes a couple of times, but I can't achieve what I want to see specifically.

          I basically need to identify unique firms which have changed their type of ownership from "okfs==31 | okfs==32 | okfs==33 | okfs==34" into "okfs==18 | okfs==21 | okfs==22 | okfs==23 | okfs==24 | okfs==27" throughout 2012-2018 period. I have generated the "which" variable in attempt to be more specific, but the difference variable gets displayed as 1 in both cases, so whether 34 has been changed to 21, or 21 has been changed to 34 - there is still the value of 1. Therefore, I cannot differentiate these observations and count the firms which had this specific change from "okfs==31 | okfs==32 | okfs==33 | okfs==34" into "okfs==18 | okfs==21 | okfs==22 | okfs==23 | okfs==24 | okfs==27".

          Maybe there is no solution to this, but I was hoping that maybe there is something that could have been figured out. If you have any suggestions, please share them~

          Thank you in advance!

          Comment

          • #6
            I can't comment easily on what code you tried as you don't show any. But in principle once you've got your coarsened classification you can look at which firms changed. The data example is boosted by singleton panels with just one year of data and thus no scope for change. But the other firms in the data example don't change at all.


            Code:
            . gen which = cond(inrange(okfs, 31, 34), 1, cond(inlist(okfs, 21, 22, 23, 24, 27), 2, 3))

* first year  
. bysort id (year) : gen start = which[1]

* last year 
. by id: gen stop = which[_N]

* tag singleton panels 
. by id: gen singleton = _N == 1 

. * we want to count firms not observations 
. egen tag = tag(id)

. * did any firms change? 
. tab start stop if tag 

           |               stop
     start |         1          2          3 |     Total
-----------+---------------------------------+----------
         1 |        22          0          0 |        22 
         2 |         0         71          0 |        71 
         3 |         0          0          1 |         1 
-----------+---------------------------------+----------
     Total |        22         71          1 |        94 

.* ignore singletons 
. tab start stop if tag & !singleton 

           |               stop
     start |         1          2          3 |     Total
-----------+---------------------------------+----------
         1 |        17          0          0 |        17 
         2 |         0         41          0 |        41 
         3 |         0          0          1 |         1 
-----------+---------------------------------+----------
     Total |        17         41          1 |        59
            • 1 like

            Comment

            • #7
              Originally posted by Nick Cox View Post
              I can't comment easily on what code you tried as you don't show any. But in principle once you've got your coarsened classification you can look at which firms changed. The data example is boosted by singleton panels with just one year of data and thus no scope for change. But the other firms in the data example don't change at all.


              Code:
              . gen which = cond(inrange(okfs, 31, 34), 1, cond(inlist(okfs, 21, 22, 23, 24, 27), 2, 3))

* first year
. bysort id (year) : gen start = which[1]

* last year
. by id: gen stop = which[_N]

* tag singleton panels
. by id: gen singleton = _N == 1

. * we want to count firms not observations
. egen tag = tag(id)

. * did any firms change?
. tab start stop if tag

| stop
start | 1 2 3 | Total
-----------+---------------------------------+----------
1 | 22 0 0 | 22
2 | 0 71 0 | 71
3 | 0 0 1 | 1
-----------+---------------------------------+----------
Total | 22 71 1 | 94

.* ignore singletons
. tab start stop if tag & !singleton

| stop
start | 1 2 3 | Total
-----------+---------------------------------+----------
1 | 17 0 0 | 17
2 | 0 41 0 | 41
3 | 0 0 1 | 1
-----------+---------------------------------+----------
Total | 17 41 1 | 59

              This works just right!

              Thank you so much, Nick!

              Comment

              Previous Next
              Working...
              X