* This file creates new variables and 
* generate one of the 2 datafiles that are used
* in the empirical analysis: data2.dta
************************************************

*capture log close
log using data2, replace
set trace off
set more off
capture drop _all
set mem 100m
*set virtual on
set matsize 100



********************************************************
*********************************************************
* FIRST USE ALL PLANTS IN LRD
********************************************************
*********************************************************

********************************************************
* 1990
*********************************************************
use data0
drop smsaold fst cou
g newind = real(ind)
drop ind
rename newind ind

g ind2 = int(ind/100)
g ind3 = int(ind/10)

sort smsa80 
egen sumpw1 = sum(pw), by(smsa)
egen sumte1 = sum(te), by(smsa)

sort smsa80 ind2
egen sumpw2 = sum(pw), by(smsa ind2)
egen sumte2 = sum(te), by(smsa ind2)

sort smsa80 ind3
egen sumpw3 = sum(pw), by(smsa ind3)
egen sumte3 = sum(te), by(smsa ind3)

replace sumpw1 = sumpw1 - sumpw2
replace sumpw2 = sumpw2 - sumpw3
replace sumpw3 = sumpw3 -    pw

replace sumte1 = sumte1 - sumte2
replace sumte2 = sumte2 - sumte3
replace sumte3 = sumte3 -    te

g share1 = 1 - (sumpw1/sumte1)
g share2 = 1 - (sumpw2/sumte2)
g share3 = 1 - (sumpw3/sumte3)

summ
g h=1
keep ppn share* sum*
sort ppn
save tmp, replace




***********************************************
* 1980
***********************************************
use D:/drive-e/moretti2/lrd/gs82extract
g ind2 = int(ind/100)
g ind3 = int(ind/10)
rename smsa smsa80

sort smsa80 
egen sumpw1 = sum(pw), by(smsa)
egen sumte1 = sum(te), by(smsa)

sort smsa80 ind2
egen sumpw2 = sum(pw), by(smsa ind2)
egen sumte2 = sum(te), by(smsa ind2)

sort smsa80 ind3
egen sumpw3 = sum(pw), by(smsa ind3)
egen sumte3 = sum(te), by(smsa ind3)

replace sumpw1 = sumpw1 - sumpw2
replace sumpw2 = sumpw2 - sumpw3
replace sumpw3 = sumpw3 -    pw

replace sumte1 = sumte1 - sumte2
replace sumte2 = sumte2 - sumte3
replace sumte3 = sumte3 -    te

g share180 = 1- (sumpw1/sumte1)
g share280 = 1- (sumpw2/sumte2)
g share380 = 1- (sumpw3/sumte3)

summ
g h80 =1
keep ppn share* h80
sort ppn
merge ppn using tmp
drop _merge
sort ppn
save data2, replace
summ



********************************************************
*********************************************************
* SECOND, I USE ONLY PLANTS IN MY SAMPLE
********************************************************
*********************************************************

use data

sort smsa80 
egen sumpw1 = sum(pw), by(smsa)
egen sumte1 = sum(te), by(smsa)
egen sumpw180 = sum(pw80), by(smsa)
egen sumte180 = sum(te80), by(smsa)

sort smsa80 ind2
egen sumpw2 = sum(pw), by(smsa ind2)
egen sumte2 = sum(te), by(smsa ind2)
egen sumpw280 = sum(pw80), by(smsa ind2)
egen sumte280 = sum(te80), by(smsa ind2)

sort smsa80 ind3
egen sumpw3 = sum(pw), by(smsa ind3)
egen sumte3 = sum(te), by(smsa ind3)
egen sumpw380 = sum(pw80), by(smsa ind3)
egen sumte380 = sum(te80), by(smsa ind3)

replace sumpw1 = sumpw1 - sumpw2
replace sumpw2 = sumpw2 - sumpw3
replace sumpw3 = sumpw3 -    pw
replace sumte1 = sumte1 - sumte2
replace sumte2 = sumte2 - sumte3
replace sumte3 = sumte3 -    te
g share1 = 1 - (sumpw1/sumte1)
g share2 = 1 - (sumpw2/sumte2)
g share3 = 1 - (sumpw3/sumte3)

replace sumpw180 = sumpw180 - sumpw280
replace sumpw280 = sumpw280 - sumpw380
replace sumpw380 = sumpw380 -    pw80
replace sumte180 = sumte180 - sumte280
replace sumte280 = sumte280 - sumte380
replace sumte380 = sumte380 -    te80
g share180 = 1 - (sumpw180/sumte180)
g share280 = 1 - (sumpw280/sumte280)
g share380 = 1 - (sumpw380/sumte380)

drop _merge
keep ppn share* sum*
summ
sort ppn
save data2b, replace