/* This is step one in database creation from CD*/

/*
This file takes the raw data from StatTransfer and compresses it into a useable 10 digit form
The basic compress operation sums across ports of entry (which are not transferred when I do 
the StatTransfer operation, which explains why a particular HS code-country combination may have multiple
records).
*/

capture program drop shrink
prog define shrink
use "/var/data/b1clmir/Airplanes/Trade_Data/imp_detl`1'.dta", clear
* convert units variables to numeric, then save label definitions
encode unit_qy1, gen(unit1)
label save unit1 using unit1`1', replace
encode unit_qy2, gen(unit2)
label save unit2 using unit2`1', replace
* sum across all values, mean across units
collapse (sum) con_qy1_yr-air_cha_yr (mean) unit1-unit2, by(commodity cty_code)
* In the unlikely event that there is any variation across ports in units, round the mean unit to nearest integer
replace unit1 = round(unit1)
replace unit1 = round(unit1)
recast byte unit1 unit2
* re-label units
run unit1`1'
run unit2`1'
label values unit1 unit1
label values unit2 unit2
merge commodity using "/var/data/b1clmir/Airplanes/Trade_Data/concord`1'.dta"
save "/var/data/b1clmir/Airplanes/Trade_Data/hs10imports`1'.dta", replace
end


clear
set mem 600m
/*
shrink	1990
shrink	1991
shrink	1992
shrink	1993
shrink	1994
shrink	1995
shrink	1996
shrink	1997
shrink	1998
shrink	1999
shrink	2000
shrink	2001
shrink	2002
shrink	2003