st: Pair-variables and creating a new dataset

[Myriam Nourry <myriam_nourry@yahoo.fr>]

Hello Statalist, 
 
I am a new user of Stata and I am wondering if I can
do what I think whith Stata. 
 
I have a panel dataset with 74 countries between
1960-1990 and a variable (sulfur dioxid emissions per
capita, say x). My dataset looks like :
 
year             countrycode             x
1960                 1                  .
.                    .                  .
.                    .                  .
1990                 .                  .
1960                 74                 .
 .                   .                  .
 .                   .                  .
1990                 .                  .
 
I want to analyse in terms of stationarity all the
pairs (difference of emissions between country i and
j): xi-xj (in fact 2701 pairs).  With the following
program, I succeed to compute all the pairs: 
 
sort year
by year: gen freq=_N
expand freq
sort year countrycode
 
by year countrycode: gen num=_n
 
by year: gen pair_id=countrycode[freq*num]
by year: gen x2=x[freq*num]
 
drop if countrycode== pair_id
drop if countrycode>pair_id
 
by year: gen dist=x-x2
sort countrycode pair_id year
 
But, I obtain a unique variable named dist. My dataset
now looks like :
 
year     countrycode      x    pair_id       dist
1960           1         .        2         x1-x2
.              .         .        .
.              .         .        .
1990           .                  .
1960           1         .        3         x1-x3
 .             .         .        .
 .             .                  .
1990           .                  .
 
I want to test unit root (use of dickey-fuller test)
on all the 2701 pairs. I don't know if it is possible
to do this with the variable dist or if i have to
obtain 2701 variables (each corresponding to a pair)
and  then use the syntax "foreach". So, I need your
help ! 
 
I try to compute all the pair-variables with the
following program: 
 
forvalues i=1/74 {
	forvalues j=2/74 {
		if (`i'~=`j' & `i'<`j') {
		gen paire`i'`j'=dist if countrycode==`i' &
pair_id==`j'
		}
	}
}
 
but I obtain a variable paireij with 83731
observations, with only 31 non missing observations. I
would like to take into account only the non-missing
observations and obtain the following dataset : 
 
year  pair12   pair13   .    .   pair7374
1960  x1-x2    x1-x3             x73- x74             
.       .        .                                   
.       .        .                   .
1990    .        .                   .
 
Is it possible to do this with Stata and must I create
this dataset on Excel and then input it on Stata ? 
 
Thank you very much for all idea you give me.
 
Best Regards.
 
Myriam Nourry




      _____________________________________________________________________________ 
Ne gardez plus qu'une seule adresse mail ! Copiez vos mails vers Yahoo! Mail 
*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/