Generating new variable which estimates the mean of a group of identifiers

[satya otil]

Dear all,

I am currently working with a dataset which contains the following variables: id (industry identifier), year, LABSH(labour share of total output), CAPSH(capital share of total output). Only one country is analysed, so the id variable identifies the different industries within the country. An example below:


* Example generated by -dataex-. To install: ssc install dataex
clear
input float id int year float(LABSH CAPSH)
1 1970 .6649457 .33505425
2 1970 .8320854 .1679146
3 1970 .883529 .11647099
4 1970 .7319602 .26803976
5 1970 .5540994 .44590065
6 1970 .56377 .43623
7 1970 .5137289 .4862711
8 1970 .7054214 .29457858
9 1970 .719725 .28027505
10 1970 .8148143 .18518573
11 1970 .547351 .452649
12 1970 .8217311 .17826892
13 1970 .7935457 .20645434
14 1970 .8322504 .16774955
15 1970 .8353265 .16467354
16 1970 .6881071 .3118929
17 1970 .8362749 .16372514
18 1970 .6861746 .3138255
19 1970 .6176726 .3823274
20 1970 .03929371 .9607063
21 1970 .6221473 .3778527
22 1970 .6551881 .3448119
23 1970 .4139394 .5860606
24 1970 .7277795 .2722205
25 1970 .28088248 .7191175
26 1970 .8978723 .1021277
27 1970 .7667561 .23324393
28 1970 .7906137 .20938635
29 1970 .6575865 .3424135
30 1970 .6504946 .3495053
31 1970 .317168 .682832
32 1970 .23677467 .7632253
33 1970 .7566398 .24336024
34 1970 .7530184 .24698158
35 1970 .7793154 .22068456
36 1970 .7456722 .25432774
37 1970 .707044 .292956
38 1970 1 0
39 1970 . .
40 1970 .6433307 .3566693
1 1971 .6551088 .3448912
2 1971 .8298256 .17017433
3 1971 .8506684 .14933157
4 1971 .7580634 .2419366
5 1971 .5831652 .4168348
6 1971 .5422989 .4577011
7 1971 .4926429 .5073571
8 1971 .6536347 .3463653
9 1971 .6888313 .3111687
10 1971 .797429 .20257105
11 1971 .55436313 .4456368
12 1971 .7926531 .2073469
13 1971 .7290537 .27094632
14 1971 .843337 .15666297
15 1971 .846294 .153706
16 1971 .6781476 .32185245
17 1971 .8504494 .14955057
18 1971 .7019526 .29804745
19 1971 .6258004 .3741996
20 1971 .04327277 .9567272
21 1971 .6251914 .3748086
22 1971 .6310546 .36894545
23 1971 .4701412 .52985877
24 1971 .7109644 .28903553
25 1971 .28638515 .7136149
26 1971 .9259716 .0740284
27 1971 .769419 .230581
28 1971 .8131477 .18685228
29 1971 .6702957 .3297043
30 1971 .6490846 .3509154
31 1971 .3177648 .6822352
32 1971 .2364985 .7635015
33 1971 .7299294 .27007055
34 1971 .7410805 .25891954
35 1971 .7751924 .2248076
36 1971 .7362803 .26371977
37 1971 .7044449 .29555508
38 1971 1 0
39 1971 . .
40 1971 .6386511 .3613489
1 1972 .6996611 .30033895
2 1972 .8293197 .17068033
3 1972 .842303 .157697
4 1972 .7473305 .25266945
5 1972 .56066453 .4393355
6 1972 .53954387 .4604561
7 1972 .4857557 .51424426
8 1972 .6646692 .33533075
9 1972 .6807057 .3192943
10 1972 .7881901 .21180987
11 1972 .54051524 .4594848
12 1972 .7903811 .20961893
13 1972 .7504753 .2495247
14 1972 .8569328 .1430672
15 1972 .8515332 .14846681
16 1972 .6595253 .3404747
17 1972 .8586392 .1413608
18 1972 .7091877 .29081225
19 1972 .6304913 .3695087
20 1972 .04974724 .9502528
end
[/CODE]

What I would like to do is generate the average aggregate labour share and capital share for a group of industries together. For example, I want to estimate the average labour share for a group which contains the industries 27,28,29,30, and another group which contains the industries 23 and 25 only. I am aware of the mean function to estimate averages, but I do not fully understand how I could apply it in this situation. Any help on this would be greatly appreciated.

I thank you in advance.

Best,

Satya

[Clyde Schechter]

You just need to create a new variable that identifies the groups first.

gen group = 1 if inlist(id, 27, 28, 29, 30)
replace group = 2 if inlist(id, 23, 5)

by group year, sort: egen wanted = mean(LABSH) if !missing(group)

I assumed that you meant to create a variable that is the mean for each group of industries separately in each year. If that's not what you wanted, just take the reference to year out of the -by group...- command.

[satya otil]

Dear Clyde,

Thank you for your reply. As always, the code works neatly.

Best,

Satya