Creating a new variable based on all the following values of an existing variable

[Andrew Bernal]

Hello -- I was unsure how to word the title, but hopefully it offers some clue as to what I am trying to do. I have the following dataset:

* Example generated by -dataex-. To install: ssc install dataex
clear
input float year str8(k i)
1996 "DEU" "ARG"
1996 "DEU" "AUS"
1996 "DEU" "AUT"
1996 "DEU" "BEL"
1996 "DEU" "BGR"
1996 "USA" "ARG"
1996 "USA" "AUS"
1996 "USA" "AUT"
1996 "USA" "BEL"
1996 "USA" "BGR"
end

I would like to create a new variable "j" which would would allow each possible i, j pair to be represented, with respect to k. I.e. k = DEU, i = ARG, j = AUS would be the first observation, then k = DEU, i = ARG, j = AUT would be next, and so on. Once all the i, j pairs are created for DEU, the i, j pairs would then be created for USA. My initial (naive) attempt at this was:

gen j = i[_n+1]

, but of course that does not create all of the pairs for each i country that I needed. Thank you for your help.

[Clyde Schechter]

I can't say I feel confident I understand what you want. But is it, perhaps, this?

* Example generated by -dataex-. To install: ssc install dataex
clear
input float year str8(k i)
1996 "DEU" "ARG"
1996 "DEU" "AUS"
1996 "DEU" "AUT"
1996 "DEU" "BEL"
1996 "DEU" "BGR"
1996 "USA" "ARG"
1996 "USA" "AUS"
1996 "USA" "AUT"
1996 "USA" "BEL"
1996 "USA" "BGR"
end

preserve
tempfile copy
rename i j
save `copy'

restore
joinby year k using `copy'
drop if i >= j

sort year k i j

[Andrew Bernal]

Clyde, sorry for not being quite clear; but indeed, the "joinby" command was the missing piece that I needed. Thank you!