add 107

107/paper.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2041794b3f9a716c71d0e965d69dad971ec9bd464e10f2586f1f600f06be99ff
+size 1302997

107/replication_package/AER_UPLOADED/MAIN.do

@@ -0,0 +1,107 @@
+global main ~/projects/patents_and_density/AER
+
+
+* Read COMETS data 
+include $main/data/create_COMETS_Patent_ExtractForEnrico.do
+clear all
+
+* Read APAT data
+include $main/data/read_apat
+clear  all
+
+* Read Input-Output Tables
+include $main/data/read1
+clear  all
+
+* Take Comets data and apat data and creates new variables
+include $main/data_3.do
+clear all
+
+* Figure 2 -- Top panel
+include $main/kodak_stock.do
+clear all
+
+
+* Table 2 (col 1 to 4) Table 4 (col 1 to 4); Fig 2 (bottom panel) and Fig 3
+include $main/reg3.do
+clear all
+include $main/reg27.do
+clear all
+
+
+* Table 2 (col 5) Table 4 (col 5)
+include $main/data/read
+clear
+include $main/reg22.do
+clear all
+
+
+*  Table 3, Table 6 (rows 1-8), 8 (top panel); Fig 5, Fig A1
+include $main/reg.do
+clear all
+
+*  Fig 6
+include $main/reg23.do
+clear all
+
+
+* Table 1, 10, 11, 12 (column1); Appendix Table A1; Fig 1 and 4
+include $main/reg2.do
+clear all
+
+
+* Table 12 (column2)
+include $main/reg29.do
+clear all
+
+* Creates iv data for regressions
+include $main/iv_new.do
+clear all
+
+
+* Table 5
+* Note: F-stat are in Main.log (llok for XXX FTEST)
+include $main/reg11.do
+clear all
+include $main/reg13.do
+clear all
+
+
+* Tab 6 (col 9,10); tab 7; tab8 (panel b); appendix tab A3, A5, A8
+include $main/data2/citations_received.do
+clear all
+include $main/data2/citations_made.do
+clear all
+include $main/data2/density_team.do
+clear all
+include $main/data2/density14_W.do
+clear all
+include $main/data2/density_14_3_reshaped.do 
+clear all
+include $main/reg20.do
+clear all
+
+
+* Appendix Table A2
+include $main/reg5.do
+clear all
+
+* Appendix Table A4
+include $main/reg36.do
+clear all
+
+* Appendix Table A6
+include $main/reg31.do
+clear all
+include $main/reg30.do
+clear all
+
+
+* Appendix Table A7
+include $main/data2/data_4.do
+clear all
+include $main/reg4.do
+clear all
+include $main/reg34.do
+clear all
+

107/replication_package/AER_UPLOADED/data/Use_SUT_Framework_2007_2012_DET.csv

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:102eec0c0330af065d8e482d8f3ea726c79c820e5838675233e750b89aec0b96
+size 303309

107/replication_package/AER_UPLOADED/data/apat63_99.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:021a4a445368162093dd79b46088c78daf3e6f87837d106b58c542651dd9bb63
+size 236903179

107/replication_package/AER_UPLOADED/data/bea_codes.dta

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5b41064edb79e3f41ad2395e2c5f5b5be756fe033495682b128f2c8baa73c6ea
+size 12217

107/replication_package/AER_UPLOADED/data/bea_county_crosswalk.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7d9f425f23efae534d1ffcf0e7d5436b6b56eddbab10bfb152e31dfc11d68535
+size 192318

107/replication_package/AER_UPLOADED/data/create_COMETS_Patent_ExtractForEnrico.do

@@ -0,0 +1,83 @@
+*************************************************************************
+* Dan Wilson
+* Program creates COMETS Patent dataset for use in Dan Wilson's Biotech Tax Rate project
+*************************************************************************
+*global DATE = "091715"	// Specify date to use in filenames of created data sets
+version 13.1
+capture log close
+*log using "/data11/home/public/dan/biotech-taxrate/data/log_files/create_COMETS_Patent.$S_DATE.log", replace
+clear
+set more off
+pause on
+
+
+//Prepare patent citation counts data
+use $main/data/patent_cite_counts, clear
+collapse (sum) citations, by(patent_id)
+sum citations, detail
+save  $main/data/cite_counts, replace
+
+
+//Prepare patent class data
+use $main/data/patent_us_classes, clear
+keep if pos==1  //keep only first listed tech class for each patent
+gen class = substr(us_class, 1, strpos(us_class,"/") - 1)
+destring class, replace force
+drop pos us_class
+save  $main/data/us_classes, replace
+
+
+//Prepare organization-specific data
+use $main/data/patent_assignees if org_id~="", clear
+egen org_namemode = mode(org_name), minmode by(org_id)
+egen org_norm_namemode = mode(org_norm_name), minmode by(org_id)
+egen org_typemode = mode(org_type), minmode by(org_id)
+egen org_state = mode(state), minmode by(org_id)
+egen org_country = mode(country), minmode by(org_id)
+drop org_name org_norm_name org_type
+rename org_namemode org_name
+rename org_norm_namemode org_norm_name
+rename org_typemode org_type
+egen tag=tag(org_id)
+keep if tag
+keep org_id org_name org_type org_norm_name org_state org_country
+save $main/data/org_id_info, replace
+*/
+
+///Create COMETS_patent dataset
+use $main/data/patent_inventors, clear
+keep patent_id last_name first_name middle_name bea_code city state postal_code country app_date
+replace country="US" if country=="USA"
+keep if country=="US"|country==""
+drop country
+gen year=substr(app_date,1,4)
+destring year, replace
+drop if year==.
+order year
+
+// Merge with data sets containing assignees, zd categories, USPTO classes, and citation counts
+*rename state state_inventor
+merge m:m patent_id using $main/data/patent_assignees, keepusing(org_id) keep(1 3) nogen
+*rename state org_state
+*rename country org_country
+*rename state_inventor state
+merge m:m patent_id using $main/data/patent_zd_cats, keepusing(zd) keep(1 3) nogenerate
+merge m:1 patent_id using $main/data/cite_counts, keepusing(citations) keep(1 3) nogenerate
+replace citations = 0 if citations==.  //citations==. means that patent had no citations
+merge m:1 patent_id using $main/data/us_classes, keepusing(class) keep(1 3) nogenerate
+
+// Create single variable for inventor name
+egen inventor=concat(first_name middle_name last_name), punct(" ")
+replace inventor=subinstr(inventor,";","",.)
+replace inventor=subinstr(inventor,"/","",.)
+*replace inventor=subinstr(inventor,".","",.)
+replace inventor=subinstr(inventor,".","",.) if strpos(inventor,".")==1
+*replace inventor=subinstr(inventor,"'","",.)
+replace inventor=itrim(inventor)
+replace inventor=ltrim(inventor)
+replace inventor=rtrim(inventor)
+drop last_name first_name middle_name 
+
+merge m:1 org_id using $main/data/org_id_info, keepusing(org_name org_type org_norm_name org_state org_country) keep(1 3) nogen
+save $main/data/COMETS_Patent_ExtractForEnrico, replace
+

107/replication_package/AER_UPLOADED/data/documenation/CodebookCOMETS20.xlsx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ead056f4e093a1c6320f45a28474f3c46de30b2a66aae664efc541f54c165ccc
+size 48771

107/replication_package/AER_UPLOADED/data/documenation/apat63_99_documentation.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:06acc94e285428837ebda59cf2a60cf2ffab39536e4137521ba6f6530e2a4e7e
+size 3420

107/replication_package/AER_UPLOADED/data/kodak_stock.csv

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:42e29d13d4ebd3e91c8403216172c82d230840fd5706801b9507c6dfc15720f6
+size 597

107/replication_package/AER_UPLOADED/data/patent_assignees.dta

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:80a8abdd1dbf65ec6536639b86d9ec06c9b664f1989767b9b65df781b6a2c010
+size 2292960713

107/replication_package/AER_UPLOADED/data/patent_citations.dta

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ec22b40ddf79cee48450876ee1b6d977580b395abcf454098c29e90477b02df9
+size 482882892

107/replication_package/AER_UPLOADED/data/patent_cite_counts.dta

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61fe03684a1fb42067bf50449aa4d09e6d698c87afb8e366088adfaea4c6bc55
+size 162059609

107/replication_package/AER_UPLOADED/data/patent_inventors.dta

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a21a824f4bcc65062c56f14da52c92915e44fa67a8eef6722b1b20297beebead
+size 5027517574

107/replication_package/AER_UPLOADED/data/patent_us_classes.dta

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8aeeb4c35e13b3e3401b2f323a183c2bb1994ff478f2b42fcaddfa41b94fac21
+size 434142421

107/replication_package/AER_UPLOADED/data/patent_zd_cats.dta

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5e4d7dda8dbbb2cf6a0b54151ca73471d74b65360711e2b98bb9fa6ebff00113
+size 45697023

107/replication_package/AER_UPLOADED/data/read.do

@@ -0,0 +1,23 @@
+u $main/data/tva1
+
+keep fips pop90 vhouse90  medfaminc90 trade90 agr90  const9 manuf90 emp90 white90
+keep if pop ~=.
+sort fips
+save tmp, replace
+
+
+clear
+insheet using $main/data/bea_county_crosswalk.txt
+rename ea bea_code
+keep fips bea
+sort fips
+merge m:1 fips using tmp
+keep if _merge ==3
+drop _merge
+
+g nonwhite90 = 1 - white90 
+collapse vhouse90  medfaminc90 trade90 agr90  const9 manuf90 nonwhite90 (sum) pop emp, by(bea)
+summ
+save $main/data/read, replace
+!rm $main/tmp.dta
+

107/replication_package/AER_UPLOADED/data/read1.do

@@ -0,0 +1,48 @@
+
+* Read Input-Output data  
+* Keep NAICS 333316	Photographic and photocopying equipment manufacturing
+insheet using $main/data/Use_SUT_Framework_2007_2012_DET.csv
+keep v1 v2 v92
+rename v1 naics
+rename v2 industry
+destring v92, replace
+keep if v92 ~=. & v92 >0 
+
+egen sum = sum(v92)
+g share = v92/sum
+rename v92 value
+drop sum 
+
+sort naics
+save $main/data/Use_SUT_Framework_2007_2012_DET, replace
+summ
+clear
+
+
+
+* Cross-walk from patents class  to industry NAICS
+* It's not one to one. It has assocated probabilities
+insheet using $main/data/uspc_Class_to_naics07_6.txt  
+rename naics naics
+rename uspc class
+
+
+* merge the two datasets
+tostring naics, replace
+sort naics
+merge m:1 naics using $main/data/Use_SUT_Framework_2007_2012_DET
+tab _merge
+drop if _merge ==2
+drop _merge
+
+* each class has multiple naics and attached probabilities. Here I use probabilities 
+* to compute the expected value of inputs used  by the photography industry
+replace value = 0 if value ==.
+collapse value [weight =probability], by(class)
+
+* Save
+sort class
+save $main/data/read1, replace
+summ
+
+

107/replication_package/AER_UPLOADED/data/read_apat.do

@@ -0,0 +1,6 @@
+insheet  using $main/data/apat63_99.txt
+rename patent patent_id
+sort patent_id
+save $main/data/apat63_99, replace
+summ
+

107/replication_package/AER_UPLOADED/data/tva1.dta

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:144a6366591e8fa3edca57b00f177531a0ce0852de01b2c883869ce5bee565c3
+size 68433611

107/replication_package/AER_UPLOADED/data/uspc_Class_to_naics07_6.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b1cd979fcee2cb79bc6946bb05581159f5ed8d10e84a108d622a79edc2c46ffa
+size 72146

107/replication_package/AER_UPLOADED/data2/citations_made.do

@@ -0,0 +1,91 @@
+
+*****************************************************
+* First I assign a bea and zd to each cited patent
+*****************************************************
+u $main/data2/data_patent_level_3
+keep patent_id zd2 bea 
+rename patent_id cite_to_patent
+rename bea  bea_cited_to
+rename zd2  zd_cited_to
+sort cite_to_patent
+save $main/data2/tmp, replace
+
+
+u $main/data/patent_citations.dta
+keep cite_to_patent cite_from_patent
+sort  cite_to_patent
+merge m:m cite_to_patent using $main/data2/tmp
+drop if _merge ==2
+drop _merge
+save $main/data2/tmp2, replace
+
+
+
+*****************************************************
+* Now I assign a bea  to each citing patent
+*****************************************************
+u $main/data2/data_patent_level_3
+keep patent_id zd2 bea
+rename patent_id cite_from_patent
+rename bea  bea_cited_from
+rename zd2  zd_cited_from
+sort cite_from_patent
+save $main/data2/tmp3, replace
+
+
+u $main/data2/tmp2
+sort  cite_from_patent
+merge m:m cite_from_patent using $main/data2/tmp3
+drop if _merge ==2
+drop _merge
+
+g same_bea = (bea_cited_from == bea_cited_to)
+g same_zd = (zd_cited_from == zd_cited_to)
+g same_bea_zd = (zd_cited_from == zd_cited_to & bea_cited_from == bea_cited_to)
+g same_bea_diff_zd = (zd_cited_from ~= zd_cited_to & bea_cited_from == bea_cited_to)
+summ
+
+collapse same_bea same_zd same_bea_zd  same_bea_diff_zd (count) citations_made = cite_to_patent, by(cite_from_patent)
+
+rename cite_from_patent patent_id
+sort patent_id
+summ
+save $main/data2/tmp4, replace
+clear
+
+
+*****************************************************
+*****************************************************
+******************************************************
+******************************************************
+u $main/data2/data_patent_level_3
+sort patent_id
+merge m:1 patent_id using $main/data2/tmp4
+tab _merge
+drop _merge
+
+
+* In cases where there are multiple cities, zd or classes within 1 year, pick mode
+* I use  maxmode to deal with ties. Later I will try minmode to see if it makes a difference
+egen main_bea        = mode(bea),     maxmode by(inventor_id year)
+egen main_zd         = mode(zd2),     maxmode by(inventor_id year)
+
+
+* Collapse by inventor_id year. Note that main_org main_org_type are constant within a
+* inventor_id year pair. I include them in by() so that they are in the 
+* collapsed data
+summ
+sort inventor_id year  
+collapse main_bea main_zd same_bea same_zd same_bea_zd  same_bea_diff_zd (sum) citations_made , by(inventor_id year)
+
+
+rename main_bea bea_code
+rename main_zd zd2
+
+drop if inventor ==.
+
+save $main/data2/citations_made, replace
+summ
+
+
+

107/replication_package/AER_UPLOADED/data2/citations_received.do

@@ -0,0 +1,81 @@
+
+*****************************************************
+* First I assign a bea to cited patents
+*****************************************************
+u $main/data2/data_patent_level_3
+keep patent_id bea 
+rename patent_id cite_to_patent
+rename bea  bea_cited_to
+sort cite_to_patent
+save $main/data2/tmp, replace
+
+
+u $main/data/patent_citations.dta
+keep cite_to_patent cite_from_patent
+sort  cite_to_patent
+merge m:m cite_to_patent using $main/data2/tmp
+drop if _merge ==2
+drop _merge
+save $main/data2/tmp2, replace
+
+
+
+*****************************************************
+* Now I assign a bea  to citing patents
+*****************************************************
+u $main/data2/data_patent_level_3
+keep patent_id  bea
+rename patent_id cite_from_patent
+rename bea  bea_cited_from
+sort cite_from_patent
+save $main/data2/tmp3, replace
+
+
+u $main/data2/tmp2
+sort  cite_from_patent
+merge m:m cite_from_patent using $main/data2/tmp3
+drop if _merge ==2
+drop _merge
+
+g same_bea = (bea_cited_from == bea_cited_to)
+summ
+
+collapse same_bea , by(cite_to_patent)
+
+rename cite_to_patent patent_id
+sort patent_id
+summ
+save $main/data2/tmp4, replace
+clear
+
+
+*****************************************************
+*****************************************************
+******************************************************
+******************************************************
+u $main/data2/data_patent_level_3
+sort patent_id
+merge m:1 patent_id using $main/data2/tmp4
+tab _merge
+drop _merge
+
+
+* In cases where there are multiple cities, pick mode
+* I use  maxmode to deal with ties. Later I will try minmode to see if it makes a difference
+egen main_bea        = mode(bea),     maxmode by(inventor_id year)
+
+
+* Collapse by inventor_id year. 
+summ
+sort inventor_id year  
+collapse main_bea same_bea , by(inventor_id year)
+
+rename main_bea bea_code
+drop if inventor ==.
+
+save $main/data2/citations_received, replace
+summ
+
+
+
+

107/replication_package/AER_UPLOADED/data2/data_4.do

@@ -0,0 +1,358 @@
+*****************************************************
+*****************************************************
+*****************************************************
+*****************************************************
+* 1 MONTH, 2 MONTHS, 3 MONTHS, 6 MONTHS, 2 YEARS, 3 YEARS
+******************************************************
+******************************************************
+*****************************************************
+*****************************************************
+
+
+*****************************************************
+*****************************************************
+* 1 MONTH
+******************************************************
+******************************************************
+u $main/data2/data_patent_level_3
+
+* When a patent is joint, I split it among inventors
+egen tmp7 = count(inventor_id), by(patent_id) 
+summ tmp7, detail
+tab tmp7 if tmp7 <=20
+g number = 1/tmp7
+replace citations = citations / tmp7
+drop tmp
+
+* MONTH
+g month =  substr(app_date,6,2)
+tab month
+
+* MODE
+egen main_bea        = mode(bea),     maxmode by(inventor_id year month)
+egen main_zd         = mode(zd2),     maxmode by(inventor_id year month)
+egen main_class      = mode(class),   maxmode by(inventor_id year month)
+egen main_org        = mode(org_id),  maxmode by(inventor_id year month)
+
+* COLLAPSE
+summ
+sort inventor_id year month
+collapse main_bea main_zd main_class general original (sum) number citat, by(inventor_id year month main_org  )
+
+rename main_bea bea_code
+rename main_zd zd2
+rename main_class class
+rename main_org org_new
+
+
+* MERGE
+sort bea zd year
+merge m:1 bea zd year using $main/data2/density14_3
+tab _merge
+drop _merge
+
+* SAVE
+g Den_bea_zd = density14 
+drop density* 
+drop if inventor ==.
+
+save $main/data2/data_4a, replace
+summ
+
+
+
+
+*****************************************************
+*****************************************************
+* 2 MONTHS
+******************************************************
+******************************************************
+u $main/data2/data_patent_level_3
+
+* When a patent is joint, I split it among inventors
+egen tmp7 = count(inventor_id), by(patent_id)
+summ tmp7, detail
+tab tmp7 if tmp7 <=20
+g number = 1/tmp7
+replace citations = citations / tmp7
+drop tmp
+
+* MONTH
+g month =  substr(app_date,6,2)
+g       tmp = 1 if month=="01"  | month=="02"  
+replace tmp = 2 if month=="03"  | month=="04"  
+replace tmp = 3 if month=="05"  | month=="06"  
+replace tmp = 4 if month=="07" | month=="08" 
+replace tmp = 5 if month=="09" | month=="10"
+replace tmp = 6 if month=="11" | month=="12"
+drop month
+g month = tmp
+tab month
+
+* MODE
+egen main_bea        = mode(bea),     maxmode by(inventor_id year month)
+egen main_zd         = mode(zd2),     maxmode by(inventor_id year month)
+egen main_class      = mode(class),   maxmode by(inventor_id year month)
+egen main_org        = mode(org_id),  maxmode by(inventor_id year month)        
+
+* COLLAPSE
+summ
+sort inventor_id year month
+collapse main_bea main_zd main_class general original (sum) number citat, by(inventor_id year month main_org)
+
+rename main_bea bea_code
+rename main_zd zd2
+rename main_class class
+rename main_org org_new
+
+
+* MERGE
+sort bea zd year
+merge m:1 bea zd year using $main/data2/density14_3
+tab _merge
+drop _merge
+
+* SAVE
+g Den_bea_zd = density14
+drop density*
+drop if inventor ==.
+
+save $main/data2/data_4e, replace
+summ
+
+
+*****************************************************
+*****************************************************
+* 3 MONTHS
+******************************************************
+******************************************************
+u $main/data2/data_patent_level_3
+
+* When a patent is joint, I split it among inventors
+egen tmp7 = count(inventor_id), by(patent_id)
+summ tmp7, detail
+tab tmp7 if tmp7 <=20
+g number = 1/tmp7
+replace citations = citations / tmp7
+drop tmp
+
+* MONTH
+g month =  substr(app_date,6,2)
+g       tmp = 1 if month=="01"  | month=="02"  | month=="03" 
+replace tmp = 2 if month=="04"  | month=="05"  | month=="06"
+replace tmp = 3 if month=="07"  | month=="08"  | month=="09"
+replace tmp = 4 if month=="10" | month=="11" | month=="12"
+drop month
+g month = tmp
+drop tmp
+tab month
+
+* MODE
+egen main_bea        = mode(bea),     maxmode by(inventor_id year month)
+egen main_zd         = mode(zd2),     maxmode by(inventor_id year month)
+egen main_class      = mode(class),   maxmode by(inventor_id year month)
+egen main_org        = mode(org_id),  maxmode by(inventor_id year month)
+
+* COLLAPSE
+summ
+sort inventor_id year month
+collapse main_bea main_zd main_class general original (sum) number citat, by(inventor_id year month main_org)
+
+rename main_bea bea_code
+rename main_zd zd2
+rename main_class class
+rename main_org org_new
+
+
+* MERGE
+sort bea zd year
+merge m:1 bea zd year using $main/data2/density14_3
+tab _merge
+drop _merge
+
+* SAVE
+g Den_bea_zd = density14
+drop density*
+drop if inventor ==.
+
+save $main/data2/data_4f, replace
+summ
+
+
+
+*****************************************************
+*****************************************************
+* 6 MONTH
+******************************************************
+******************************************************
+u $main/data2/data_patent_level_3
+
+* When a patent is joint, I split it among inventors
+egen tmp7 = count(inventor_id), by(patent_id)
+summ tmp7, detail
+tab tmp7 if tmp7 <=20
+g number = 1/tmp7
+replace citations = citations / tmp7
+drop tmp
+
+
+* MONTH
+g month1 =  substr(app_date,6,2)
+g       tmp = 1 if month=="01"  | month=="02"
+replace tmp = 1 if month=="03"  | month=="04"
+replace tmp = 1 if month=="05"  | month=="06"
+replace tmp = 2 if month=="07"  | month=="08"
+replace tmp = 2 if month=="09"  | month=="10"
+replace tmp = 2 if month=="11" | month=="12"
+g month = tmp
+drop tmp month1
+tab month
+
+
+* MODE
+egen main_bea        = mode(bea),     maxmode by(inventor_id year month)
+egen main_zd         = mode(zd2),     maxmode by(inventor_id year month)
+egen main_class      = mode(class),   maxmode by(inventor_id year month)
+egen main_org        = mode(org_id),  maxmode by(inventor_id year month)
+
+* COLLAPSE
+summ
+sort inventor_id year month
+collapse main_bea main_zd main_class general original (sum) number citat, by(inventor_id year month main_org)
+
+rename main_bea bea_code
+rename main_zd zd2
+rename main_class class
+rename main_org org_new
+
+
+* MERGE
+sort bea zd year
+merge m:1 bea zd year using $main/data2/density14_3
+tab _merge
+drop _merge
+
+* SAVE
+g Den_bea_zd = density14
+drop density* 
+drop if inventor ==.
+
+save $main/data2/data_4b, replace
+summ
+
+
+
+
+
+*****************************************************
+*****************************************************
+* 2 YEARS
+******************************************************
+******************************************************
+u $main/data2/data_patent_level_3
+
+* When a patent is joint, I split it among inventors
+egen tmp7 = count(inventor_id), by(patent_id) 
+summ tmp7, detail
+tab tmp7 if tmp7 <=20
+g number = 1/tmp7
+replace citations = citations / tmp7
+drop tmp
+
+
+g tmp = int(year/2) * 2
+drop year
+rename tmp year
+tab year
+
+
+* MODE
+egen main_bea        = mode(bea),     maxmode by(inventor_id year )
+egen main_zd         = mode(zd2),     maxmode by(inventor_id year )
+egen main_class      = mode(class),   maxmode by(inventor_id year )
+egen main_org        = mode(org_id),  maxmode by(inventor_id year )
+
+* COLLAPSE
+summ
+sort inventor_id year 
+collapse main_bea main_zd main_class general original (sum) number citat, by(inventor_id year main_org)
+
+rename main_bea bea_code
+rename main_zd zd2
+rename main_class class
+rename main_org org_new
+
+
+* MERGE
+sort bea zd year
+merge m:1 bea zd year using $main/data2/density14_3
+tab _merge
+drop _merge
+
+* SAVE
+g Den_bea_zd = density14 
+drop density* 
+drop if inventor ==.
+
+save $main/data2/data_4c, replace
+summ
+
+
+
+
+
+
+*****************************************************
+*****************************************************
+* 3 YEARS
+******************************************************
+******************************************************
+u $main/data2/data_patent_level_3
+
+* When a patent is joint, I split it among inventors
+egen tmp7 = count(inventor_id), by(patent_id)
+summ tmp7, detail
+tab tmp7 if tmp7 <=20
+g number = 1/tmp7
+replace citations = citations / tmp7
+drop tmp
+
+
+g tmp = int(year/3) * 3
+drop year
+rename tmp year
+tab year
+
+
+* MODE
+egen main_bea        = mode(bea),     maxmode by(inventor_id year )
+egen main_zd         = mode(zd2),     maxmode by(inventor_id year )
+egen main_class      = mode(class),   maxmode by(inventor_id year )
+egen main_org        = mode(org_id),  maxmode by(inventor_id year )
+
+* COLLAPSE
+summ
+sort inventor_id year
+collapse main_bea main_zd main_class general original (sum) number citat, by(inventor_id year main_org)
+
+rename main_bea bea_code
+rename main_zd zd2
+rename main_class class
+rename main_org org_new
+
+
+* MERGE
+sort bea zd year
+merge m:1 bea zd year using $main/data2/density14_3
+tab _merge
+drop _merge
+
+* SAVE
+g Den_bea_zd = density14
+drop density* 
+drop if inventor ==.
+
+save $main/data2/data_4d, replace
+summ
+
+

107/replication_package/AER_UPLOADED/data2/density14_W.do

@@ -0,0 +1,177 @@
+********************************************************************
+********************************************************************
+* Measures of cluster size that put some weight on qualty of
+* inventors (as opposed just their raw number)
+*******************************************************************
+********************************************************************
+
+
+
+***********************************************
+***********************************************
+* CLEAN THE DATA
+***********************************************
+***********************************************
+u $main/data/COMETS_Patent_ExtractForEnrico
+
+keep if year >1970 & year <=2007
+egen inventor_id = group(inventor)
+drop if inventor_id ==.
+drop if class ==.
+
+drop if bea =="NOCOUNTRY"
+drop if bea =="OTHER_USA"
+drop if bea =="UNITED STATES"
+destring bea, replace
+drop if bea ==.
+drop if bea ==0
+
+* ZD is the broad technology class
+* Here I create a ZD2 which is the numeric version 
+egen zd2 = group(zd)
+tab zd zd2
+drop if zd2 ==.
+
+
+* Cases of  patents with multiple inventors
+*egen N = count(year), by(inventor_id patent_id)
+*summ N
+*drop N
+
+* MERGE WITH NBER PATENT DATA
+sort patent
+merge m:m patent using $main/data/apat63_99
+tab _merge
+summ
+drop _merge
+drop if inventor_id ==.
+
+
+
+****************************************************************
+*****************************************************************
+* DATA CHECK:
+* MULTIPLE CITIES AND TECHNOLOGY PER YEAR
+****************************************************************
+*****************************************************************
+* I drop if number of different cities or zd in a given year too large
+g N = 1
+sort inventor year bea
+by inventor year: replace N = N[_n-1] + 1 if bea ~= bea[_n-1] & _n >1
+egen NN = max(N), by(inventor)
+drop if NN > 3 & NN ~=.
+drop N NN
+
+g N = 1
+sort inventor year zd2
+by inventor year: replace N = N[_n-1] + 1 if zd2 ~= zd2[_n-1] & _n >1
+egen NN = max(N), by(inventor year)
+drop if NN > 3 & NN ~=.
+drop N NN
+
+
+***********************************************
+***************************************************
+* MEASURES OF CLUSTERS DENSITY
+* WEIGHTED
+***************************************************
+***************************************************
+drop zd
+
+
+* number of lifetime citations and patents, by inventor
+egen M = sum(citations), by(inventor_id )
+egen MM = count(patent_id), by(inventor_id)
+* The data are at the patent level. I want to count inventors
+sort inventor_id year bea zd
+by inventor_id year bea zd: g d1 = 1 if _n ==1
+keep if d1 ==1
+
+
+summ M MM, detail
+* K's and H's are indicators if an inventor has lifetime patents or citations 
+* above a thereshold
+g K = 0
+g K2 = 0
+g K3 = 0
+g H = 0
+g H2 = 0
+g H3 = 0
+
+replace K =  d1 if M >= 5
+replace  K2 = d1 if M >= 10
+replace  K3 = d1 if M >= 20
+
+replace H =  d1 if MM >= 3
+replace  H2 = d1 if MM >= 5
+replace  H3 = d1 if MM >= 10
+
+egen count = sum(K), by(bea zd year)
+egen total = sum(K), by(zd year)
+g density14_W1 = (count-K)/(total-K)
+drop count total 
+
+egen count = sum(K2), by(bea zd year)
+egen total = sum(K2), by(zd year)
+g density14_W2 = (count-K2)/(total-K2)
+drop count total
+
+egen count = sum(K3), by(bea zd year)
+egen total = sum(K3), by(zd year)
+g density14_W3 = (count-K3)/(total-K3)
+drop count total
+
+
+egen count = sum(H), by(bea zd year)
+egen total = sum(H), by(zd year)
+g density14_WW1 = (count-H)/(total-H)
+drop count total
+
+egen count = sum(H2), by(bea zd year)
+egen total = sum(H2), by(zd year)
+g density14_WW2 = (count-H2)/(total-H2)
+drop count total
+
+egen count = sum(H3), by(bea zd year)
+egen total = sum(H3), by(zd year)
+g density14_WW3 = (count-H3)/(total-H3)
+drop count total
+
+
+g Den_bea_zd_W1 = density14_W1
+g Den_bea_zd_W2 = density14_W2
+g Den_bea_zd_W3 = density14_W3
+g Den_bea_zd_WW1 = density14_WW1
+g Den_bea_zd_WW2 = density14_WW2
+g Den_bea_zd_WW3 = density14_WW3
+drop density14_W*
+
+
+***********************************************
+* WEIGHTS
+***********************************************
+egen count = sum(M), by(bea zd year)
+egen total = sum(M), by(zd year)
+g density14_W4 = (count-M)/(total-M)
+*replace density14_W4 = 0 if density14_W4 <0
+drop count total
+
+egen count = sum(MM), by(bea zd year)
+egen total = sum(MM), by(zd year)
+g density14_WW4 = (count-MM)/(total-MM)
+*replace density14_WW4 = 0 if density14_WW4 <0
+
+collapse density14_W4 density14_WW4 Den_bea_zd_W1 Den_bea_zd_W2  Den_bea_zd_W3 Den_bea_zd_WW1 Den_bea_zd_WW2  Den_bea_zd_WW3, by(inventor_id year)
+g Den_bea_zd_W4 = density14_W4
+g Den_bea_zd_WW4 = density14_WW4
+sort  inventor_id year
+save $main/data2/density14_W, replace
+summ
+
+
+
+
+
+
+
+

107/replication_package/AER_UPLOADED/data2/density_14_3_reshaped.do

@@ -0,0 +1,9 @@
+
+use $main/data2/density14_3
+drop density14B
+rename density14 density
+reshape wide density, i(bea year) j(zd)
+summ
+sort bea year
+save  $main/data2/density14_3_reshaped, replace
+

107/replication_package/AER_UPLOADED/data2/density_team.do

@@ -0,0 +1,103 @@
+********************************************************************
+********************************************************************
+* This is like data_3, but instead of being at the inventor level it
+* is at the team level. ( Based on Referee #2 suggestion)
+*******************************************************************
+********************************************************************
+
+
+
+***********************************************
+***********************************************
+* CLEAN THE DATA
+***********************************************
+***********************************************
+u $main/data/COMETS_Patent_ExtractForEnrico
+
+keep if year >1970 & year <=2007
+egen inventor_id = group(inventor)
+drop if inventor_id ==.
+drop if class ==.
+
+drop if bea =="NOCOUNTRY"
+drop if bea =="OTHER_USA"
+drop if bea =="UNITED STATES"
+destring bea, replace
+drop if bea ==.
+drop if bea ==0
+
+* ZD is the broad technology class
+* Here I create a ZD2 which is the numeric version 
+egen zd2 = group(zd)
+tab zd zd2
+drop if zd2 ==.
+
+
+* Cases of  patents with multiple inventors
+*egen N = count(year), by(inventor_id patent_id)
+*summ N
+*drop N
+
+* MERGE WITH NBER PATENT DATA
+sort patent
+merge m:m patent using $main/data/apat63_99
+tab _merge
+summ
+drop _merge
+drop if inventor_id ==.
+
+
+
+****************************************************************
+*****************************************************************
+* DATA CHECK:
+* MULTIPLE CITIES AND TECHNOLOGY PER YEAR
+****************************************************************
+*****************************************************************
+* I drop if number of different cities or zd in a given year too large
+g N = 1
+sort inventor year bea
+by inventor year: replace N = N[_n-1] + 1 if bea ~= bea[_n-1] & _n >1
+egen NN = max(N), by(inventor)
+drop if NN > 3 & NN ~=.
+drop N NN
+
+g N = 1
+sort inventor year zd2
+by inventor year: replace N = N[_n-1] + 1 if zd2 ~= zd2[_n-1] & _n >1
+egen NN = max(N), by(inventor year)
+drop if NN > 3 & NN ~=.
+drop N NN
+
+
+***********************************************
+**************************************
+* MEASURES OF CLUSTERS DENSITY
+* I LEAVE OUT MEMBERS OF TEAM
+***************************************
+***********************************************
+
+* Team size: number of inventors on a patent
+egen team  = count(inventor_id), by(patent_id)
+summ team, detail
+egen tt  = mean(team), by(inventor_id year bea zd)
+sort inventor_id year bea zd 
+by inventor_id year bea zd: g d1 = 1 if _n ==1
+
+* this measure of densities varies across inventors depending on their mean 
+* team size 
+egen count = count(d1), by(bea zd year)
+egen total = count(d1), by(zd year)
+g density14_team = (count-tt)/(total-tt)
+
+
+sort inventor_id year
+collapse density14_team , by(inventor_id year)
+save $main/data2/density_team, replace
+summ
+
+
+
+
+
+

107/replication_package/AER_UPLOADED/data_3.do

@@ -0,0 +1,380 @@
+********************************************************************
+********************************************************************
+* This file reads the data used in the AER tax paper and creates 2 dataset 
+* a) data.dta is a balanced rectangual dataset at the inventor-year level
+* b) data_patent_level is at the patent level, like the original data
+*
+* zd is a broad technological area
+* class is a finer technology class
+* bea is a locality code: it is a consolidated metro area
+*
+* Codebook can be found here: https://www.kauffman.org/microsites/comets/codebook
+* Or google “Zucker-Darby COMETS patent data”
+* Source of data is the AER paper with Dan W. Dan made original extract 
+*******************************************************************
+********************************************************************
+
+
+
+***********************************************
+***********************************************
+* CLEAN THE DATA
+***********************************************
+***********************************************
+u $main/data/COMETS_Patent_ExtractForEnrico.dta 
+
+keep if year >1970 & year <=2007
+egen inventor_id = group(inventor)
+drop if inventor_id ==.
+drop if class ==.
+
+drop if bea =="NOCOUNTRY"
+drop if bea =="OTHER_USA"
+drop if bea =="UNITED STATES"
+destring bea, replace
+drop if bea ==.
+drop if bea ==0
+
+* ZD is the broad technology class
+* Here I create a ZD2 which is the numeric version 
+egen zd2 = group(zd)
+tab zd zd2
+drop if zd2 ==.
+
+
+* MERGE WITH NBER PATENT DATA
+sort patent
+merge m:m patent using  $main/data/apat63_99
+tab _merge
+summ
+drop _merge
+drop if inventor_id ==.
+
+
+
+*****************************************************************
+g N = 1
+sort inventor year bea
+by inventor year: replace N = N[_n-1] + 1 if bea ~= bea[_n-1] & _n >1
+egen NN = max(N), by(inventor)
+drop if NN > 3 & NN ~=.
+drop N NN
+
+g N = 1
+sort inventor year zd2
+by inventor year: replace N = N[_n-1] + 1 if zd2 ~= zd2[_n-1] & _n >1
+egen NN = max(N), by(inventor year)
+drop if NN > 3 & NN ~=.
+drop N NN
+
+
+***********************************************
+**************************************
+* MEASURES OF CLUSTERS DENSITY
+* LEAVE OUT MEANS
+* I want to measure the thickness of the labor market
+* I can measure it as (a) numbner of inventors or (2) number of firms 
+***************************************
+***********************************************
+
+* NUMBER OF INVENTORS
+***************************
+sort inventor_id year bea 
+by inventor_id year bea: g d1 = 1 if _n ==1
+egen count = count(d1), by(bea year)
+egen total = count(d1), by(year)
+g density13 = (count -1)/(total-1)
+drop count total d1
+
+sort inventor_id year bea zd 
+by inventor_id year bea zd: g d1 = 1 if _n ==1
+egen count = count(d1), by(bea zd year)
+egen total = count(d1), by(zd year)
+g density14 = (count-1)/(total-1)
+save $main/data2/data_3_count, replace
+drop count total d1
+
+sort inventor_id year bea class
+by inventor_id year bea class: g d1 = 1 if _n ==1
+egen count = count(d1), by(bea class year)
+egen total = count(d1), by(class year)
+g density15 = (count-1)/(total-1)
+drop count total d1
+
+
+
+sort inventor_id year bea cat
+by inventor_id year bea cat: g d1 = 1 if _n ==1
+egen count = count(d1), by(bea cat year)
+egen total = count(d1), by(cat year)
+g density20 = (count-1)/(total-1)
+drop count total d1
+
+sort inventor_id year bea subcat
+by inventor_id year bea subcat: g d1 = 1 if _n ==1
+egen count = count(d1), by(bea subcat year)
+egen total = count(d1), by(subcat year)
+g density21 = (count-1)/(total-1)
+drop count total d1
+
+
+
+
+* NUMBER OF FIRMS
+**********************************************
+sort org_id year bea
+by org_id year bea: g d1 = 1 if _n ==1
+egen count = count(d1), by(bea year)
+egen total = count(d1), by(year)
+g density13B = (count -1)/(total-1)
+drop count total d1
+
+sort org_id year bea zd
+by org_id year bea zd: g d1 = 1 if _n ==1
+egen count = count(d1), by(bea zd year)
+egen total = count(d1), by(zd year)
+g density14B = (count-1)/(total-1)
+drop count total d1
+
+sort org_id year bea class
+by org_id year bea class: g d1 = 1 if _n ==1
+egen count = count(d1), by(bea class year)
+egen total = count(d1), by(class year)
+g density15B = (count-1)/(total-1)
+drop count total d1
+
+
+sort org_id year bea cat
+by org_id year bea cat: g d1 = 1 if _n ==1
+egen count = count(d1), by(bea cat year)
+egen total = count(d1), by(cat year)
+g density20B = (count-1)/(total-1)
+drop count total d1
+
+sort org_id year bea subcat
+by org_id year bea subcat: g d1 = 1 if _n ==1
+egen count = count(d1), by(bea subcat year)
+egen total = count(d1), by(subcat year)
+g density21B = (count-1)/(total-1)
+drop count total d1
+
+
+
+* I SAVE DENSITY MEASURES IN DATA FILES 
+* TO BE USED LATER
+***********************************************
+save $main/data2/data_patent_level_3, replace
+summ
+
+sort bea year
+collapse density13 density13B, by(bea year)
+save $main/data2/density13_3, replace
+summ
+
+u $main/data2/data_patent_level_3
+sort bea zd2 year
+collapse density14 density14B, by(bea zd2 year)
+save $main/data2/density14_3, replace
+summ
+
+u $main/data2/data_patent_level_3
+sort bea class year
+collapse density15 density15B, by(bea class year)
+save $main/data2/density15_3, replace
+summ
+
+
+
+
+u $main/data2/data_patent_level_3
+sort bea cat year
+collapse density20 density20B, by(bea cat year)
+save $main/data2/density20_3, replace
+summ
+
+u $main/data2/data_patent_level_3
+sort bea subcat year
+collapse density21 density21B, by(bea subcat year)
+save $main/data2/density21_3, replace
+summ
+
+
+
+* ORGANIZATION
+u $main/data2/data_patent_level_3
+drop if org_id ==""
+g org_N = 1 if inventor_id ~=.
+sort org_id bea year
+collapse (sum) org_N, by(org_id org_norm_ bea year)
+summ
+save $main/data2/organizations_3, replace
+
+
+
+*****************************************************
+*****************************************************
+* I COLLAPSE 
+* THE SAMPLE BECOMES AT THE INVENTOR-YEAR LEVEL
+*
+* IF AN INVENTOR HAS MORE THAN ONE CITY OR SECTOR IN A GIVEN 
+* YEAR, I USE MODAL CITY AND SECTOR. 
+******************************************************
+******************************************************
+u $main/data2/data_patent_level_3
+
+* In cases where there are multiple cities, zd or classes within 1 year, pick mode
+* I use  maxmode to deal with ties. Minmode does not  make a difference
+egen main_bea        = mode(bea),     maxmode by(inventor_id year)
+egen main_zd         = mode(zd2),     maxmode by(inventor_id year)
+egen main_class      = mode(class),   maxmode by(inventor_id year)
+egen main_org        = mode(org_id),  maxmode by(inventor_id year)
+egen main_cat        = mode(cat),     maxmode by(inventor_id year)
+egen main_subcat     = mode(subcat),  maxmode by(inventor_id year)
+egen main_org_type   = mode(org_typ), maxmode by(inventor_id year)
+
+
+* When a patent is joint, I split it equally among inventors
+egen tmp7 = count(inventor_id), by(patent_id) 
+summ tmp7, detail
+tab tmp7 if tmp7 <=20
+g number = 1/tmp7
+replace citations = citations / tmp7
+drop tmp7
+
+* Team size: number of inventors on a patent
+egen team  = count(inventor_id), by(patent_id)
+summ team, detail
+* indicator for solo inventor
+g    team2 = (team ==1)
+summ year team*
+
+* Collapse by inventor_id year. Note that main_org main_org_type are constant within a
+* inventor_id year pair. I include them in by() so that they are in the 
+* collapsed data
+summ
+sort inventor_id year  
+collapse main_bea main_zd main_class main_cat main_subcat general original team team2 (sum) number citat, by(inventor_id year main_org main_org_type )
+
+
+
+********************************************
+********************************************
+* I FILLIN AND INTERPOLATE SMALL GAPS
+* I CREATE INDICATORS FOR INTERPOLATION THAT ALLOW ME
+* TO IDENTIFY AND DROP INTERPOLATED  OBSERVATIONS
+********************************************
+********************************************
+rename main_bea bea_code
+rename main_zd zd2
+rename main_class class
+rename main_org org_id
+rename main_org_type org_type
+rename main_cat cat
+rename main_subcat subcat
+
+fillin inventor_id year
+summ
+
+g inter_bea    =0
+g inter_zd     =0
+g inter_class  =0
+g inter_bea2    =0
+g inter_zd2     =0
+g inter_class2  =0
+
+* 1 YEAR GAP
+sort inventor year
+by inventor: replace inter_bea = 1   if bea ==. & bea[_n-1] == bea[_n+1] & bea[_n-1] ~=.
+by inventor: replace bea = bea[_n-1] if inter_bea ==1
+
+by inventor: replace inter_zd = 1   if zd ==. & zd[_n-1] == zd[_n+1] & zd[_n-1] ~=.
+by inventor: replace zd = zd[_n-1] if inter_zd ==1
+
+by inventor: replace inter_class = 1   if class ==. & class[_n-1] == class[_n+1] & class[_n-1] ~=.
+by inventor: replace class = class[_n-1] if inter_class ==1
+
+
+* 2 YEAR GAP
+by inventor: replace inter_bea2 = 1  if bea ==. & bea[_n-1] ==. & bea[_n-2] == bea[_n+1] & bea[_n-2] ~=.
+by inventor: replace bea = bea[_n-2] if inter_bea2 ==1
+
+by inventor: replace inter_zd2 = 1 if zd ==. & zd[_n-1] ==. & zd[_n-2] == zd[_n+1] & zd[_n-2] ~=.
+by inventor: replace zd = zd[_n-2] if inter_zd2 ==1
+
+by inventor: replace inter_class2 = 1 if class ==. & class[_n-1] ==. & class[_n-2] == class[_n+1] & class[_n-2] ~=.
+by inventor: replace class = class[_n-2] if inter_class2 ==1
+
+
+summ inter* 
+summ bea if inter_bea ==0
+summ bea if inter_bea ==1
+summ class if inter_class ==0
+summ class if inter_class ==1
+
+summ bea if inter_bea2 ==0
+summ bea if inter_bea2 ==1
+summ class if inter_class2 ==0
+summ class if inter_class2 ==1
+
+
+*******************************************
+*****************************
+* MERGE MAIN DATA WITH 
+* MEASURES OF CLUSTER DENSITY
+*****************************
+********************************************
+sum
+sort bea year
+merge m:1 bea year using $main/data2/density13_3
+tab _merge
+drop _merge
+
+sort bea zd year
+merge m:1 bea zd year using $main/data2/density14_3
+tab _merge
+drop _merge
+
+sort bea class year
+merge m:1 bea class year using $main/data2/density15_3
+tab _merge
+drop _merge
+
+
+sort bea cat year
+merge m:1 bea cat year using $main/data2/density20_3
+tab _merge
+drop _merge
+
+
+sort bea subcat year
+merge m:1 bea subcat year using $main/data2/density21_3
+tab _merge
+drop _merge
+
+
+*************************************
+*************************************
+* SAVE
+************************************
+**************************************
+g Den_bea = density13 
+g Den_bea_zd = density14 
+g Den_bea_class = density15 
+g Den_bea_cat    = density20
+g Den_bea_subcat = density21
+
+g Den_beaB = density13B
+g Den_bea_zdB = density14B
+g Den_bea_classB = density15B
+g Den_bea_catB    = density20B
+g Den_bea_subcatB = density21B
+
+drop density* Den_bea_cat* Den_bea_subcatB* cat subcat
+drop if inventor ==.
+
+save $main/data2/data_3, replace
+summ
+
+
+
+

107/replication_package/AER_UPLOADED/iv_new.do

@@ -0,0 +1,122 @@
+********************************************************************
+********************************************************************
+********************************************************************
+********************************************************************
+********************************************************************
+********************************************************************
+* IV 
+********************************************************************
+********************************************************************
+********************************************************************
+********************************************************************
+********************************************************************
+********************************************************************
+clear
+u $main/data2/data_3
+
+egen total = sum(number), by(inventor)
+
+* INCLUDE THE FOLLOWING 6 LINES IF YOU DON'T WANT INTERPOLATED DATA
+replace bea    =. if inter_bea   ==1
+replace zd     =. if inter_zd    ==1
+replace class  =. if inter_class ==1
+replace bea    =. if inter_bea2  ==1
+replace zd     =. if inter_zd2   ==1
+replace class  =. if inter_class2==1
+
+
+drop if year  ==.
+drop if zd ==.
+drop if org_id == ""
+drop if bea ==.
+egen org_new2 = group(org_id)
+
+* This is the number of inventors in a  firm  in a given city (in a given field and year)
+g n = 1
+collapse (sum) n, by(year zd bea org_new)
+
+* Now the level of observation is year field city firm
+* This is the total number of inventors in a city across all firms (in a given field and year)
+egen nn = sum(n), by(year zd bea)
+
+* This is the share of  inventors in a firm in  a given city (in a given field and year)
+g share = n/nn
+summ n nn share, detail
+
+* Lagged share
+sort zd bea org_new year
+by   zd bea org_new: g share_1 = share[_n-1]
+by   zd bea org_new: g n_1 = n[_n-1]
+
+
+* The data are now at the year field city firm level
+* This is the total number of inventors by firm in the city (in a given year and field)
+* It's the same as n above
+egen r2 = sum(n), by(org_new  year zd bea)
+* This is the total number of inventors in a firm across all cities in the nation (in a given year and filed)
+egen rr2 = sum(n), by(org_new  year zd )
+* For each firm, field city and year, this is the total number of inventors that the firm has 
+* in the nation outside the relevant city (in a year)
+g DD = rr2 - r2
+
+
+
+* This is the change over time in DD
+* Note: it is non-missing only for firms that are present in a given city-field in year t-1 and t
+*       If a firm is not present in year t-1, DD1 is missing
+sort zd org_new year
+by zd org_new: g DD1 = DD - DD[_n-1]
+by zd org_new: g DD2 = (DD - DD[_n-1]) / DD[_n-1]
+by zd org_new: g DD3 = (DD - DD[_n-1]+1) / (DD[_n-1] +1)
+by zd org_new: g DD4 = (DD - DD[_n-1]+0.01) / (DD[_n-1] +0.01)
+
+
+
+* First Differences models are defined 
+* Only for Obs in Contiguous years
+sort zd bea org_new year
+by zd bea org_new: g Dyear = year - year[_n-1]
+replace share_1 = . if Dyear ~=1
+replace n_1 = . if Dyear ~=1
+replace DD1 = . if Dyear ~=1
+replace DD2 = . if Dyear ~=1
+replace DD3 = . if Dyear ~=1
+
+
+* Now I normalize  DD1 by the relevant  nationwide chnage 
+g tmp8 = DD1
+egen tmp9 = sum(tmp8), by(zd year)
+g iv8 = tmp8/tmp9
+drop tmp8 tmp9
+
+
+* * THIS IS THE TOTAL IN THE FOCAL CITY, FIELD ANF YEAR
+save $main/data2/tmp1, replace
+collapse (sum) iv8 , by(year zd bea )
+summ
+sort year zd bea         
+save $main/data2/tmp3, replace
+
+
+*  THIS IS THE CHANGE IN FOCAL FIRM
+clear
+u $main/data2/tmp1
+rename iv8 hh8
+collapse hh8 , by(year zd bea org_new2)
+sort year zd bea
+
+merge m:1  year zd bea using $main/data2/tmp3
+tab _merge
+drop _merge
+
+* NOW I SUBTRACT THE CHANGE IN FOCAL FIRM FROM THE TOTAL SO THAT THE INSTRUMENT DOES NOT CONTAIN FOCAL FIRM
+replace iv8 = iv8 - hh8 if hh8 ~=.
+
+sort bea year zd org_new2
+save $main/data2/iv_data_new, replace
+summ
+
+
+
+
+

107/replication_package/AER_UPLOADED/kodak_stock.do

@@ -0,0 +1,14 @@
+* Source: Kodak: http://www.1stock1.com/1stock1_173.htm
+insheet using $main/data/kodak_stock.csv
+
+rename ending price
+list year price
+
+destring price, replace
+keep if year >=1990 & year <2010
+
+line  price year , legend(off) lpattern(solid dot  dot)  lcolor(black black black)  saving($main/tables/kodak_stock,replace)
+graph export $main/tables/kodak_stock.pdf, replace
+
+
+

107/replication_package/AER_UPLOADED/reg.do

@@ -0,0 +1,483 @@
+********************************************************************
+********************************************************************
+********************************************************************
+********************************************************************
+********************************************************************
+********************************************************************
+
+
+********************************************************************
+********************************************************************
+********************************************************************
+* READ DATA
+********************************************************************
+********************************************************************
+********************************************************************
+clear
+u $main/data2/data_3
+
+egen total = sum(number), by(inventor)
+
+* INCLUDE THE FOLLOWING 6 LINES IF YOU DON'T WANT INTERPOLATED DATA
+replace bea    =. if inter_bea   ==1
+replace zd     =. if inter_zd    ==1
+replace class  =. if inter_class ==1
+replace bea    =. if inter_bea2  ==1
+replace zd     =. if inter_zd2   ==1
+replace class  =. if inter_class2==1
+
+
+
+* CLUSTER SIZE 
+g x   = log(Den_bea_zd    )
+sort inventor year
+by inventor : g  x_m1 = x[_n-1]
+by inventor : g  x_m2 = x[_n-2]
+by inventor : g  x_m3 = x[_n-3]
+by inventor : g  x_m4 = x[_n-4]
+by inventor : g  x_m5 = x[_n-5]
+by inventor : g  x_p1 = x[_n+1]
+by inventor : g  x_p2 = x[_n+2]
+by inventor : g  x_p3 = x[_n+3]
+by inventor : g  x_p4 = x[_n+4]
+by inventor : g  x_p5 = x[_n+5]
+
+
+
+
+drop if year  ==.
+drop if zd ==.
+drop if bea ==.
+egen org_new2 = group(org_id) if org_id ~= ""
+
+
+
+********************************************************************
+********************************************************************
+********************************************************************
+********************************************************************
+********************************************************************
+********************************************************************
+
+* NUMBER OF PATENTS IN A YEAR
+g y = log(number)
+
+* CITATIONS
+* To avoid dropping the 0's, I add 0.00001
+g y2 = log(citations+0.00001)
+g y3 = log( (citations+0.00001) / number) 
+
+* SPECIALIZATION
+*g y4 = 1-general
+
+* NUMBER OF FIRMS
+g x3  = log(Den_bea_zdB) 
+
+
+***********************
+***********************
+* STAR INVENTORS
+* top 10% = 3.25
+***********************
+***********************
+keep if total  >=3.25
+
+
+
+********************************************************************
+********************************************************************
+********************************************************************
+* NEW VARIABLES
+********************************************************************
+********************************************************************
+********************************************************************
+egen cluster1               = group(bea zd)
+egen cluster0               = group(bea zd year)
+egen cluster2               = group(bea )
+egen cluster_bea_year       = group(bea year)
+egen cluster_bea_zd_year    = group(bea zd year)
+egen cluster_bea_class_year = group(bea class year)
+egen cluster_zd_year        = group(zd year)
+egen cluster_bea_class      = group(bea class)
+egen cluster_class_year     = group(class year)
+egen org_new                = group(org_id)
+egen cluster_org_year       = group(org_new year)
+egen org_size               = count(inventor), by(org_new year)
+egen lab_size              = count(inventor), by(org_new year bea)
+
+
+
+********************************************************************
+********************************************************************
+********************************************************************
+* REGRESSIONS
+********************************************************************
+********************************************************************
+********************************************************************
+keep if y~=. & zd ~=. & class ~=. & year ~=. & inventor ~=. & cluster1 ~=. & bea ~=.
+summ y y2 y3  x 
+summ number, detail
+summ x Den_bea_zd , detail
+tab zd
+
+
+***************************************
+*BASELINE
+***************************************
+
+* MAIN
+eststo: reghdfe y x , absorb(year bea zd class                                                                                        ) vce(cluster cluster1) keepsin 
+eststo: reghdfe y x , absorb(year bea zd class cluster1                                                                               ) vce(cluster cluster1) keepsin 
+eststo: reghdfe y x , absorb(year bea zd class cluster1 cluster_bea_class                                                             ) vce(cluster cluster1) keepsin 
+eststo: reghdfe y x , absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year                                             ) vce(cluster cluster1) keepsin 
+eststo: reghdfe y x , absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year                          ) vce(cluster cluster1) keepsin 
+eststo: reghdfe y x , absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor                 ) vce(cluster cluster1) keepsin 
+eststo: reghdfe y x , absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor cluster_bea_year) vce(cluster cluster1) keepsin 
+eststo: reghdfe y x , absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor cluster_bea_year org_new                                   ) vce(cluster cluster1) keepsin
+
+esttab using tables/table1.tex, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+esttab using tables/table1, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+eststo clear
+
+
+
+* NON LINEARITIES
+summ x, detail
+egen pctile25 = pctile(x), p(25) by(zd year)
+egen pctile50 = pctile(x), p(50) by(zd year)
+egen pctile75 = pctile(x), p(75) by(zd year)
+g dd1 = 0
+g dd2 = 0
+g dd3 = 0
+g dd4 = 0
+replace dd1 =1 if x                <= pctile25
+replace dd2 =1 if x > pctile25 & x <= pctile50
+replace dd3 =1 if x > pctile50 & x <= pctile75
+replace dd4 =1 if x > pctile75
+g x_dd1 = x*dd1
+g x_dd2 = x*dd2
+g x_dd3 = x*dd3
+g x_dd4 = x*dd4
+summ dd1 dd2 dd3 dd4
+
+eststo: reghdfe y x_dd1 x_dd2 x_dd3 x_dd4 , absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor cluster_bea_year) vce(cluster cluster1) keepsin
+test x_dd1 =x_dd2 =x_dd3 =x_dd4
+
+eststo: reghdfe y x_dd1 x_dd2 x_dd3 x_dd4 , absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor cluster_bea_year org_new) vce(cluster cluster1) keepsin
+test x_dd1 =x_dd2 =x_dd3 =x_dd4
+
+esttab using tables/table90.tex, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+esttab using tables/table90, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+eststo clear
+
+
+
+* CITATIONS
+eststo: reghdfe y2 x , absorb(year bea zd class                                                                                        ) vce(cluster cluster1) keepsin 
+eststo: reghdfe y2 x , absorb(year bea zd class cluster1                                                                               ) vce(cluster cluster1) keepsin 
+eststo: reghdfe y2 x , absorb(year bea zd class cluster1 cluster_bea_class                                                             ) vce(cluster cluster1) keepsin 
+eststo: reghdfe y2 x , absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year                                             ) vce(cluster cluster1) keepsin 
+eststo: reghdfe y2 x , absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year                          ) vce(cluster cluster1) keepsin 
+eststo: reghdfe y2 x , absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor                 ) vce(cluster cluster1) keepsin 
+eststo: reghdfe y2 x , absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor cluster_bea_year) vce(cluster cluster1) keepsin 
+eststo: reghdfe y2 x , absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor cluster_bea_year org_new  ) vce(cluster cluster1) keepsin
+esttab using tables/table4.tex, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+esttab using tables/table4, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+eststo clear
+
+eststo: reghdfe y3 x , absorb(year bea zd class                                                                                        ) vce(cluster cluster1) keepsin
+eststo: reghdfe y3 x , absorb(year bea zd class cluster1                                                                               ) vce(cluster cluster1) keepsin
+eststo: reghdfe y3 x , absorb(year bea zd class cluster1 cluster_bea_class                                                             ) vce(cluster cluster1) keepsin
+eststo: reghdfe y3 x , absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year                                             ) vce(cluster cluster1) keepsin
+eststo: reghdfe y3 x , absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year                          ) vce(cluster cluster1) keepsin
+eststo: reghdfe y3 x , absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor                 ) vce(cluster cluster1) keepsin
+eststo: reghdfe y3 x , absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor cluster_bea_year) vce(cluster cluster1) keepsin
+eststo: reghdfe y3 x , absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor cluster_bea_year org_new  ) vce(cluster cluster1) keepsin 
+esttab using tables/table5.tex, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+esttab using tables/table5, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+eststo clear
+
+
+
+* CLUSTER SIZE IS NUMBER OF FIRMS
+eststo: reghdfe y x3, absorb(year bea zd class                                                                                        ) vce(cluster cluster1) keepsin 
+eststo: reghdfe y x3, absorb(year bea zd class cluster1                                                                               ) vce(cluster cluster1) keepsin 
+eststo: reghdfe y x3, absorb(year bea zd class cluster1 cluster_bea_class                                                             ) vce(cluster cluster1) keepsin 
+eststo: reghdfe y x3, absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year                                             ) vce(cluster cluster1) keepsin 
+eststo: reghdfe y x3, absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year                          ) vce(cluster cluster1) keepsin 
+eststo: reghdfe y x3, absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor                 ) vce(cluster cluster1) keepsin 
+eststo: reghdfe y x3, absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor cluster_bea_year) vce(cluster cluster1) keepsin 
+eststo: reghdfe y x3, absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor cluster_bea_year org_new                                   ) vce(cluster cluster1) keepsin
+esttab using tables/table11.tex, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+esttab using tables/table11, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+eststo clear
+
+
+
+***************************************
+* STAYERS 
+***************************************
+preserve
+
+* SAME CITY
+egen max = max(bea), by(inventor)
+egen min = min(bea), by(inventor)
+keep if max == min
+drop max min
+
+eststo: reghdfe y x , absorb(year bea zd class                                                                                        ) vce(cluster cluster1) keepsin 
+eststo: reghdfe y x , absorb(year bea zd class cluster1                                                                               ) vce(cluster cluster1) keepsin 
+eststo: reghdfe y x , absorb(year bea zd class cluster1 cluster_bea_class                                                             ) vce(cluster cluster1) keepsin 
+eststo: reghdfe y x , absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year                                             ) vce(cluster cluster1) keepsin 
+eststo: reghdfe y x , absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year                          ) vce(cluster cluster1) keepsin 
+eststo: reghdfe y x , absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor                 ) vce(cluster cluster1) keepsin 
+eststo: reghdfe y x , absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor cluster_bea_year) vce(cluster cluster1) keepsin 
+eststo: reghdfe y x , absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor cluster_bea_year org_new                                   ) vce(cluster cluster1) keepsin
+esttab using tables/table7.tex, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+esttab using tables/table7, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+eststo clear
+
+
+* SAME CITY AND FIELD
+egen max = max(zd), by(inventor)
+egen min = min(zd), by(inventor)
+keep if max == min
+drop max min
+
+eststo: reghdfe y x , absorb(year bea zd class                                                                                        ) vce(cluster cluster1) keepsin
+eststo: reghdfe y x , absorb(year bea zd class cluster1                                                                               ) vce(cluster cluster1) keepsin
+eststo: reghdfe y x , absorb(year bea zd class cluster1 cluster_bea_class                                                             ) vce(cluster cluster1) keepsin
+eststo: reghdfe y x , absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year                                             ) vce(cluster cluster1) keepsin
+eststo: reghdfe y x , absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year                          ) vce(cluster cluster1) keepsin
+eststo: reghdfe y x , absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor                 ) vce(cluster cluster1) keepsin
+eststo: reghdfe y x , absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor cluster_bea_year) vce(cluster cluster1) keepsin
+eststo: reghdfe y x , absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor cluster_bea_year org_new                                   ) vce(cluster cluster1) keepsin
+esttab using tables/table71.tex, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+esttab using tables/table71, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+eststo clear
+
+
+* SAME FIELD
+restore
+preserve
+egen max = max(zd), by(inventor)
+egen min = min(zd), by(inventor)
+keep if max == min
+drop max min
+
+eststo: reghdfe y x , absorb(year bea zd class                                                                                        ) vce(cluster cluster1) keepsin
+eststo: reghdfe y x , absorb(year bea zd class cluster1                                                                               ) vce(cluster cluster1) keepsin
+eststo: reghdfe y x , absorb(year bea zd class cluster1 cluster_bea_class                                                             ) vce(cluster cluster1) keepsin
+eststo: reghdfe y x , absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year                                             ) vce(cluster cluster1) keepsin
+eststo: reghdfe y x , absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year                          ) vce(cluster cluster1) keepsin
+eststo: reghdfe y x , absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor                 ) vce(cluster cluster1) keepsin
+eststo: reghdfe y x , absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor cluster_bea_year) vce(cluster cluster1) keepsin
+eststo: reghdfe y x , absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor cluster_bea_year org_new                                   ) vce(cluster cluster1) keepsin
+esttab using tables/table72.tex, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+esttab using tables/table72, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+eststo clear
+
+
+
+***************************************
+*LEADS AND LAGS - BASELINE
+***************************************
+restore
+preserve
+eststo: reghdfe y x_p5 x_p4 x_p3 x_p2 x_p1 x x_m1 x_m2 x_m3 x_m4 x_m5 ,absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor cluster_bea_year org_new  ) vce(cluster cluster1)  
+eststo clear
+
+g b1 = _b[x_p5]
+g b2 = _b[x_p4]
+g b3 = _b[x_p3]
+g b4 = _b[x_p2]
+g b5 = _b[x_p1]
+g b6 = _b[x]
+g b7 = _b[x_m1]
+g b8 = _b[x_m2]
+g b9 = _b[x_m3]
+g b10= _b[x_m4]
+g b11= _b[x_m5]
+g se1  =_se[x_p5]
+g se2  =_se[x_p4]
+g se3  =_se[x_p3]
+g se4  =_se[x_p2]
+g se5  =_se[x_p1]
+g se6  =_se[x]
+g se7  =_se[x_m1]
+g se8  =_se[x_m2]
+g se9  =_se[x_m3]
+g se10 =_se[x_m4]
+g se11 =_se[x_m5]
+
+lincom x_p5 
+g bb1  = r(estimate)
+g sse1 = r(se)
+lincom x_p5 + x_p4 
+g bb2  = r(estimate)
+g sse2 = r(se)
+lincom x_p5 + x_p4 + x_p3 
+g bb3  = r(estimate)
+g sse3 = r(se)
+lincom  x_p5 + x_p4 + x_p3 + x_p2 
+g bb4  = r(estimate)
+g sse4 = r(se)
+lincom x_p5 + x_p4 + x_p3 + x_p2 + x_p1 
+g bb5 = r(estimate)
+g sse5 = r(se)
+lincom x_p5 + x_p4 + x_p3 + x_p2 + x_p1 + x 
+g bb6  = r(estimate)
+g sse6 = r(se)
+lincom x_p5 + x_p4 + x_p3 + x_p2 + x_p1 + x + x_m1 
+g bb7  = r(estimate)
+g sse7 = r(se)
+lincom x_p5 + x_p4 + x_p3 + x_p2 + x_p1 + x + x_m1 + x_m2 
+g bb8  = r(estimate)
+g sse8 = r(se)
+lincom x_p5 + x_p4 + x_p3 + x_p2 + x_p1 + x + x_m1 + x_m2 + x_m3 
+g bb9  = r(estimate)
+g sse9 = r(se)
+lincom x_p5 + x_p4 + x_p3 + x_p2 + x_p1 + x + x_m1 + x_m2 + x_m3 + x_m4 
+g bb10  = r(estimate)
+g sse10 = r(se)
+lincom x_p5 + x_p4 + x_p3 + x_p2 + x_p1 + x + x_m1 + x_m2 + x_m3 + x_m4 + x_m5
+g bb11  = r(estimate)
+g sse11 = r(se)
+
+keep if _n==1
+keep b1-b11 se1-se11 bb* sse*
+save beta1, replace
+
+clear
+u beta1
+g n=_n
+reshape long b se bb sse, i(n) j(j)
+g t = j-6
+g upper = b + 1.96*se
+g lower = b - 1.96*se
+
+
+line  b upper lower t , legend(off) lpattern(solid dot  dot)  lcolor(black black black)     xlabel(-5 "{&beta}{sub:5}" -4 "{&beta}{sub:4}" -3 "{&beta}{sub:3}" -2 "{&beta}{sub:2}" -1 "{&beta}{sub:1}" 0 "{&beta}{sub:0}" 1 "{&beta}{sub:-1}" 2 "{&beta}{sub:-2}" 3 "{&beta}{sub:-3}" 4 "{&beta}{sub:-4}" 5 "{&beta}{sub:-5}" ) saving(fig1,replace)
+graph export fig1.pdf, replace
+! mv -f beta1.dta tables/
+! mv -f fig1.pdf tables/
+! mv -f fig1.gph tables/
+
+drop upper lower
+g upper = bb + 1.96*sse
+g lower = bb - 1.96*sse
+
+
+line  bb upper lower t , legend(off) lpattern(solid dot  dot)  lcolor(black black black)     xlabel(-5 "{&mu}{sub:5}" -4 "{&mu}{sub:4}" -3 "{&mu}{sub:3}" -2 "{&mu}{sub:2}" -1 "{&mu}{sub:1}" 0 "{&mu}{sub:0}" 1 "{&mu}{sub:-1}" 2 "{&mu}{sub:-2}" 3 "{&mu}{sub:-3}" 4 "{&mu}{sub:-4}" 5 "{&mu}{sub:-5}" ) saving(fig1b,replace)
+graph export fig1b.pdf, replace
+! mv -f fig1b.pdf tables/
+! mv -f fig1b.gph tables/
+
+
+
+
+
+
+***************************************
+*LEADS AND LAGS - DRUGS
+***************************************
+restore
+preserve
+keep if class == 424 
+eststo: reghdfe y x_p5 x_p4 x_p3 x_p2 x_p1 x x_m1 x_m2 x_m3 x_m4 x_m5 ,absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor cluster_bea_year org_new  ) vce(cluster cluster1)
+eststo clear
+
+g b1 = _b[x_p5]
+g b2 = _b[x_p4]
+g b3 = _b[x_p3]
+g b4 = _b[x_p2]
+g b5 = _b[x_p1]
+g b6 = _b[x]
+g b7 = _b[x_m1]
+g b8 = _b[x_m2]
+g b9 = _b[x_m3]
+g b10= _b[x_m4]
+g b11= _b[x_m5]
+g se1  =_se[x_p5]
+g se2  =_se[x_p4]
+g se3  =_se[x_p3]
+g se4  =_se[x_p2]
+g se5  =_se[x_p1]
+g se6  =_se[x]
+g se7  =_se[x_m1]
+g se8  =_se[x_m2]
+g se9  =_se[x_m3]
+g se10 =_se[x_m4]
+g se11 =_se[x_m5]
+
+lincom x_p5
+g bb1  = r(estimate)
+g sse1 = r(se)
+lincom x_p5 + x_p4
+g bb2  = r(estimate)
+g sse2 = r(se)
+lincom x_p5 + x_p4 + x_p3
+g bb3  = r(estimate)
+g sse3 = r(se)
+lincom  x_p5 + x_p4 + x_p3 + x_p2
+g bb4  = r(estimate)
+g sse4 = r(se)
+lincom x_p5 + x_p4 + x_p3 + x_p2 + x_p1
+g bb5 = r(estimate)
+g sse5 = r(se)
+lincom x_p5 + x_p4 + x_p3 + x_p2 + x_p1 + x
+g bb6  = r(estimate)
+g sse6 = r(se)
+lincom x_p5 + x_p4 + x_p3 + x_p2 + x_p1 + x + x_m1
+g bb7  = r(estimate)
+g sse7 = r(se)
+lincom x_p5 + x_p4 + x_p3 + x_p2 + x_p1 + x + x_m1 + x_m2
+g bb8  = r(estimate)
+g sse8 = r(se)
+lincom x_p5 + x_p4 + x_p3 + x_p2 + x_p1 + x + x_m1 + x_m2 + x_m3
+g bb9  = r(estimate)
+g sse9 = r(se)
+lincom x_p5 + x_p4 + x_p3 + x_p2 + x_p1 + x + x_m1 + x_m2 + x_m3 + x_m4
+g bb10  = r(estimate)
+g sse10 = r(se)
+lincom x_p5 + x_p4 + x_p3 + x_p2 + x_p1 + x + x_m1 + x_m2 + x_m3 + x_m4 + x_m5
+g bb11  = r(estimate)
+g sse11 = r(se)
+
+
+
+keep if _n==1
+keep b1-b11 se1-se11 bb* sse*
+save beta2, replace
+
+clear
+u beta2
+g n=_n
+reshape long b se bb sse, i(n) j(j)
+g t = j-6
+g upper = b + 1.96*se
+g lower = b - 1.96*se
+
+
+line  b upper lower t , legend(off) lpattern(solid dot  dot)  lcolor(black black black)     xlabel(-5 "{&beta}{sub:5}" -4 "{&beta}{sub:4}" -3 "{&beta}{sub:3}" -2 "{&beta}{sub:2}" -1 "{&beta}{sub:1}" 0 "{&beta}{sub:0}" 1 "{&beta}{sub:-1}" 2 "{&beta}{sub:-2}" 3 "{&beta}{sub:-3}" 4 "{&beta}{sub:-4}" 5 "{&beta}{sub:-5}" ) saving(fig2,replace)
+graph export fig2.pdf, replace
+! mv -f beta2.dta tables/
+! mv -f fig2.pdf tables/
+! mv -f fig2.gph tables/
+
+drop upper lower
+g upper = bb + 1.96*sse
+g lower = bb - 1.96*sse
+
+
+line  bb upper lower t , legend(off) lpattern(solid dot  dot)  lcolor(black black black)     xlabel(-5 "{&mu}{sub:5}" -4 "{&mu}{sub:4}" -3 "{&mu}{sub:3}" -2 "{&mu}{sub:2}" -1 "{&mu}{sub:1}" 0 "{&mu}{sub:0}" 1 "{&mu}{sub:-1}" 2 "{&mu}{sub:-2}" 3 "{&mu}{sub:-3}" 4 "{&mu}{sub:-4}" 5 "{&mu}{sub:-5}" ) saving(fig2b,replace)
+graph export fig2b.pdf, replace
+! mv -f fig2b.pdf tables/
+! mv -f fig2b.gph tables/
+
+
+
+
+restore
+*THE PART BELOW WAS MOVED TO REG23.DO
+

107/replication_package/AER_UPLOADED/reg11.do

@@ -0,0 +1,242 @@
+********************************************************************
+********************************************************************
+********************************************************************
+* IV --  BASELINE  
+********************************************************************
+********************************************************************
+********************************************************************
+
+
+********************************************************************
+********************************************************************
+********************************************************************
+* READ DATA
+********************************************************************
+********************************************************************
+********************************************************************
+clear
+u $main/data2/data_3
+
+egen total = sum(number), by(inventor)
+
+* INCLUDE THE FOLLOWING 6 LINES IF YOU DON'T WANT INTERPOLATED DATA
+replace bea    =. if inter_bea   ==1
+replace zd     =. if inter_zd    ==1
+replace class  =. if inter_class ==1
+replace bea    =. if inter_bea2  ==1
+replace zd     =. if inter_zd2   ==1
+replace class  =. if inter_class2==1
+
+
+
+* CLUSTER SIZE 
+g x   = log(Den_bea_zd    )
+sort inventor year
+by inventor : g  x_m1 = x[_n-1]
+by inventor : g  x_m2 = x[_n-2]
+by inventor : g  x_m3 = x[_n-3]
+by inventor : g  x_m4 = x[_n-4]
+by inventor : g  x_m5 = x[_n-5]
+by inventor : g  x_p1 = x[_n+1]
+by inventor : g  x_p2 = x[_n+2]
+by inventor : g  x_p3 = x[_n+3]
+by inventor : g  x_p4 = x[_n+4]
+by inventor : g  x_p5 = x[_n+5]
+
+
+
+
+drop if year  ==.
+drop if zd ==.
+drop if bea ==.
+egen org_new2 = group(org_id) if org_id ~= ""
+
+
+
+********************************************************************
+********************************************************************
+********************************************************************
+********************************************************************
+********************************************************************
+********************************************************************
+
+* NUMBER OF PATENTS IN A YEAR
+g y = log(number)
+
+* CITATIONS
+* To avoid dropping the 0's, I add 0.00001
+g y2 = log(citations+0.00001)
+g y3 = log( (citations+0.00001) / number) 
+
+* SPECIALIZATION
+*g y4 = 1-general
+
+* NUMBER OF FIRMS
+g x3  = log(Den_bea_zdB) 
+
+
+***********************
+***********************
+* DEFINE STAR INVENTORS
+***********************
+***********************
+keep if total  >=3.25
+
+
+
+********************************************************************
+********************************************************************
+********************************************************************
+* NEW VARIABLES
+********************************************************************
+********************************************************************
+********************************************************************
+egen cluster1               = group(bea zd)
+egen cluster0               = group(bea zd year)
+egen cluster2               = group(bea )
+egen cluster_bea_year       = group(bea year)
+egen cluster_bea_zd_year    = group(bea zd year)
+egen cluster_bea_class_year = group(bea class year)
+egen cluster_zd_year        = group(zd year)
+egen cluster_bea_class      = group(bea class)
+egen cluster_class_year     = group(class year)
+egen org_new                = group(org_id)
+egen cluster_org_year       = group(org_new year)
+egen org_size               = count(inventor), by(org_new year)
+egen lab_size              = count(inventor), by(org_new year bea)
+
+
+
+********************************************************************
+********************************************************************
+********************************************************************
+* REGRESSIONS
+********************************************************************
+********************************************************************
+********************************************************************
+keep if y~=. & zd ~=. & class ~=. & year ~=. & inventor ~=. & cluster1 ~=. & bea ~=.
+summ y y2 y3  x 
+summ number, detail
+summ x Den_bea_zd , detail
+tab zd
+
+
+***************************************
+*MODELS IN DIFFERENCES and IV
+***************************************
+preserve
+
+
+drop if org_id == ""
+sort org_new2 bea year zd
+merge m:1   bea year zd org_new2 using  $main/data2/iv_data_new
+drop _merge
+g IV = iv8
+
+sort inventor year
+by inventor: g Dy    = y - y[_n-1]
+by inventor: g Dx    = x - x[_n-1]
+by inventor: g Dyear = year - year[_n-1]
+by inventor: g Dbea  = 1 if bea_code == bea_code[_n-1] 
+by inventor: g Dy_2    = y - y[_n-2]
+by inventor: g Dx_2    = x - x[_n-2]
+by inventor: g Dyear_2 = year - year[_n-2]
+by inventor: g Dbea_2  = 1 if bea_code == bea_code[_n-2]
+by inventor: g Dx_m1 = x[_n-1] - x[_n-2]
+by inventor: g Dx_m2 = x[_n-2] - x[_n-3]
+by inventor: g Dx_m3 = x[_n-3] - x[_n-2]
+egen cluster10       = group(bea )
+
+
+* Models in first differences should only include observations in contiguous years
+keep if Dyear ==1
+keep if Dy ~=. & Dx ~=. & IV ~=.
+
+
+
+
+* MODELS WITH LAGS: THEY CAN BE COMPARED TO f.e. MODELS BUT CAN'T USE IV's
+eststo: reghdfe  Dy Dx Dx_m1 Dx_m2 Dx_m3 ,        absorb(year  )  vce(cluster cluster10)
+lincom Dx + Dx_m1 + Dx_m2 + Dx_m3
+eststo: reghdfe  Dy Dx Dx_m1 Dx_m2 Dx_m3 ,            absorb(year zd )  vce(cluster cluster10)
+lincom Dx + Dx_m1 + Dx_m2 + Dx_m3
+eststo: reghdfe  Dy Dx Dx_m1 Dx_m2 Dx_m3 ,            absorb(year zd class )  vce(cluster cluster10)
+lincom Dx + Dx_m1 + Dx_m2 + Dx_m3
+eststo: reghdfe  Dy Dx Dx_m1 Dx_m2 Dx_m3 ,            absorb(year zd class org_new)  vce(cluster cluster10)
+lincom Dx + Dx_m1 + Dx_m2 + Dx_m3
+eststo: reghdfe  Dy Dx Dx_m1 Dx_m2 Dx_m3 ,            absorb(year zd class org_new cluster_zd_year )  vce(cluster cluster10)
+lincom Dx + Dx_m1 + Dx_m2 + Dx_m3
+eststo: reghdfe  Dy Dx Dx_m1 Dx_m2 Dx_m3 ,            absorb(year zd class org_new cluster_zd_year cluster_class_year )  vce(cluster cluster10)
+lincom Dx + Dx_m1 + Dx_m2 + Dx_m3
+esttab using tables11/table29.tex, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+esttab using tables11/table29, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+eststo clear
+
+
+
+* OLS
+eststo: reghdfe  Dy Dx, absorb(year                  )  vce(cluster cluster10)
+eststo: reghdfe  Dy Dx, absorb(year  zd              )  vce(cluster cluster10)
+eststo: reghdfe  Dy Dx, absorb(year  zd class        )  vce(cluster cluster10)
+eststo: reghdfe  Dy Dx, absorb(year  zd class org_new)  vce(cluster cluster10)
+eststo: reghdfe  Dy Dx, absorb(year  zd class cluster_zd_year                    org_new)  vce(cluster cluster10)
+eststo: reghdfe  Dy Dx, absorb(year  zd class cluster_zd_year cluster_class_year org_new)  vce(cluster cluster10)
+
+esttab using tables11/table20.tex, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+esttab using tables11/table20, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+eststo clear
+
+* 2SLS
+eststo: ivreghdfe Dy (Dx =IV), absorb(year                   ) cluster(cluster10) 
+eststo: ivreghdfe Dy (Dx =IV), absorb(year   zd              ) cluster(cluster10)
+eststo: ivreghdfe Dy (Dx =IV), absorb(year   zd class        ) cluster(cluster10)
+eststo: ivreghdfe Dy (Dx =IV), absorb(year zd class org_new) cluster(cluster10)
+eststo: ivreghdfe Dy (Dx =IV), absorb(year zd class cluster_zd_year org_new ) cluster(cluster10)
+eststo: ivreghdfe Dy (Dx =IV), absorb(year zd class cluster_zd_year cluster_class_year org_new) cluster(cluster10)
+
+esttab using tables11/table22.tex, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+esttab using tables11/table22, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+eststo clear
+
+* First Stage
+eststo: reghdfe Dx IV, absorb(year                 ) vce(cluster cluster10) res(v1)
+test IV=0
+eststo: reghdfe Dx IV, absorb(year zd              ) vce(cluster cluster10) res(v2)
+test IV=0
+eststo: reghdfe Dx IV, absorb(year zd class        ) vce(cluster cluster10) res(v3)
+test IV=0
+eststo: reghdfe Dx IV, absorb(year zd class org_new) vce(cluster cluster10) res(v4)
+test IV=0
+eststo: reghdfe Dx IV, absorb(year zd class cluster_zd_year org_new) vce(cluster cluster10) res(v5) 
+test IV=0
+eststo: reghdfe Dx IV, absorb(year zd class cluster_zd_year cluster_class_year org_new) vce(cluster cluster10) res(v6)
+test IV=0
+
+esttab using tables11/table21.tex, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+esttab using tables11/table21, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+eststo clear
+
+
+* EXOGENEITY TESTS
+* XXX FSTAT
+reghdfe  Dy Dx v1, absorb(year                  )  vce(cluster cluster10)
+test v1
+reghdfe  Dy Dx v2, absorb(year  zd              )  vce(cluster cluster10)
+test v2
+reghdfe  Dy Dx v3, absorb(year  zd class        )  vce(cluster cluster10)
+test v3
+reghdfe  Dy Dx v4, absorb(year  zd class org_new)  vce(cluster cluster10)
+test v4
+reghdfe  Dy Dx v5, absorb(year  zd class cluster_zd_year                    org_new)  vce(cluster cluster10)
+test v5
+reghdfe  Dy Dx v6, absorb(year  zd class cluster_zd_year cluster_class_year org_new)  vce(cluster cluster10)
+test v6
+drop v1 v2 v3 v4 v5 v6
+
+
+restore
+
+
+
+
+

107/replication_package/AER_UPLOADED/reg13.do

@@ -0,0 +1,211 @@
+********************************************************************
+********************************************************************
+********************************************************************
+*  IV 
+********************************************************************
+********************************************************************
+********************************************************************
+
+
+********************************************************************
+********************************************************************
+********************************************************************
+* READ DATA
+********************************************************************
+********************************************************************
+********************************************************************
+clear
+u $main/data2/data_3
+
+egen total = sum(number), by(inventor)
+
+* INCLUDE THE FOLLOWING 6 LINES IF YOU DON'T WANT INTERPOLATED DATA
+replace bea    =. if inter_bea   ==1
+replace zd     =. if inter_zd    ==1
+replace class  =. if inter_class ==1
+replace bea    =. if inter_bea2  ==1
+replace zd     =. if inter_zd2   ==1
+replace class  =. if inter_class2==1
+
+
+
+* CLUSTER SIZE 
+g x   = log(Den_bea_zd    )
+sort inventor year
+by inventor : g  x_m1 = x[_n-1]
+by inventor : g  x_m2 = x[_n-2]
+by inventor : g  x_m3 = x[_n-3]
+by inventor : g  x_m4 = x[_n-4]
+by inventor : g  x_m5 = x[_n-5]
+by inventor : g  x_p1 = x[_n+1]
+by inventor : g  x_p2 = x[_n+2]
+by inventor : g  x_p3 = x[_n+3]
+by inventor : g  x_p4 = x[_n+4]
+by inventor : g  x_p5 = x[_n+5]
+
+
+
+
+drop if year  ==.
+drop if zd ==.
+drop if bea ==.
+egen org_new2 = group(org_id) if org_id ~= ""
+
+
+
+********************************************************************
+********************************************************************
+********************************************************************
+********************************************************************
+********************************************************************
+********************************************************************
+
+* NUMBER OF PATENTS IN A YEAR
+g y = log(number)
+
+* CITATIONS
+* To avoid dropping the 0's, I add 0.00001
+g y2 = log(citations+0.00001)
+g y3 = log( (citations+0.00001) / number) 
+
+* SPECIALIZATION
+*g y4 = 1-general
+
+* NUMBER OF FIRMS
+g x3  = log(Den_bea_zdB) 
+
+
+***********************
+***********************
+* DEFINE STAR INVENTORS
+***********************
+***********************
+keep if total  >=3.25
+
+
+
+********************************************************************
+********************************************************************
+********************************************************************
+* NEW VARIABLES
+********************************************************************
+********************************************************************
+********************************************************************
+egen cluster1               = group(bea zd)
+egen cluster0               = group(bea zd year)
+egen cluster2               = group(bea )
+egen cluster_bea_year       = group(bea year)
+egen cluster_bea_zd_year    = group(bea zd year)
+egen cluster_bea_class_year = group(bea class year)
+egen cluster_zd_year        = group(zd year)
+egen cluster_bea_class      = group(bea class)
+egen cluster_class_year     = group(class year)
+egen org_new                = group(org_id)
+egen cluster_org_year       = group(org_new year)
+egen org_size               = count(inventor), by(org_new year)
+egen lab_size              = count(inventor), by(org_new year bea)
+
+
+
+********************************************************************
+********************************************************************
+********************************************************************
+* REGRESSIONS
+********************************************************************
+********************************************************************
+********************************************************************
+keep if y~=. & zd ~=. & class ~=. & year ~=. & inventor ~=. & cluster1 ~=. & bea ~=.
+summ y y2 y3  x 
+summ number, detail
+summ x Den_bea_zd , detail
+tab zd
+
+
+***************************************
+*MODELS IN DIFFERENCES and IV
+***************************************
+preserve
+
+
+drop if org_id == ""
+sort org_new2 bea year zd
+merge m:1   bea year zd org_new2 using  $main/data2/iv_data_new
+drop _merge
+g IV = iv8
+
+sort inventor year
+by inventor: g Dy    = y - y[_n-1]
+by inventor: g Dx    = x - x[_n-1]
+by inventor: g Dyear = year - year[_n-1]
+by inventor: g Dbea  = 1 if bea_code == bea_code[_n-1] 
+by inventor: g Dy_2    = y - y[_n-2]
+by inventor: g Dx_2    = x - x[_n-2]
+by inventor: g Dyear_2 = year - year[_n-2]
+by inventor: g Dbea_2  = 1 if bea_code == bea_code[_n-2]
+by inventor: g Dx_m1 = x[_n-1] - x[_n-2]
+by inventor: g Dx_m2 = x[_n-2] - x[_n-3]
+by inventor: g Dx_m3 = x[_n-3] - x[_n-2]
+egen cluster10       = group(bea )
+
+
+* Models in first differences should only include observations in contiguous years
+keep if Dyear ==1
+keep if Dy ~=. & Dx ~=. & IV ~=.
+
+
+
+*******************************************
+sort inventor year
+*by inventor: g Dy_m1 = Dy[_n-1]
+*by inventor: g Dy_m2 = Dy[_n-2]
+*by inventor: g Dy_m3 = Dy[_n-3]
+
+*by inventor: g Dx_m1 = Dx[_n-1]
+*by inventor: g Dx_m2 = Dx[_n-2]
+*by inventor: g Dx_m3 = Dx[_n-3]
+
+by inventor: g IV_p1 = IV[_n+1]
+by inventor: g IV_p2 = IV[_n+2]
+by inventor: g IV_p3 = IV[_n+3]
+
+
+summ year Dy Dx IV*
+
+
+* IV
+eststo: ivreghdfe Dy (Dx =IV_p1), absorb(year zd class cluster_zd_year cluster_class_year org_new) cluster(cluster10)
+
+*eststo: ivreghdfe Dy (Dx =IV_p1 IV_p2), absorb(year zd class cluster_zd_year cluster_class_year org_new) cluster(cluster10)
+
+
+* First Stage
+eststo: reghdfe Dx IV_p1, absorb(year zd class cluster_zd_year cluster_class_year org_new) vce(cluster cluster10) res(v6)
+
+*eststo: reghdfe Dx IV_p1 IV_p2, absorb(year zd class cluster_zd_year cluster_class_year org_new) vce(cluster cluster10) res(v6)
+
+eststo: reghdfe Dx IV IV_p1, absorb(year zd class cluster_zd_year cluster_class_year org_new) vce(cluster cluster10) res(v88)
+test IV=0
+
+*eststo: reghdfe Dx IV IV_p1 IV_p2, absorb(year zd class cluster_zd_year cluster_class_year org_new) vce(cluster cluster10) res(v6)
+*test IV=0
+
+
+
+restore
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+

107/replication_package/AER_UPLOADED/reg2.do

@@ -0,0 +1,333 @@
+
+********************************************************************
+********************************************************************
+********************************************************************
+* RANKINGS, SUMMARY STATS, BINSCATTER  and COUNTERFACTUALS
+********************************************************************
+********************************************************************
+********************************************************************
+
+
+********************************************************************
+********************************************************************
+******************************
+* TOP PATENTERS
+******************************
+********************************************************************
+********************************************************************
+u $main/data2/organizations_3.dta
+collapse (sum) org_N, by(org_norm)
+gsort -org_N
+g n = _n
+keep if n <= 25
+list
+
+outfile  org_norm org_N using tables2/table4.txt, replace
+outsheet org_norm org_N using tables2/table4.csv, replace comma
+
+
+
+********************************************************************
+********************************************************************
+******************************
+* CLUSTER SIZE 
+******************************
+********************************************************************
+********************************************************************
+clear
+u $main/data2/density14_3
+
+* Some city-industry are not in the data because they have no patents. 
+* Thus, I am missing some of the 0's
+* Here I fillin do that I have every combinatiuon of city zd and year
+* and I add the 0's
+fillin bea_code zd2 year
+replace density14  =0 if density14 ==. 
+replace density14B =0 if density14B ==.
+
+
+* Add zd name and bea name
+g       zd_name = "Biology and Chemestry" if zd2 ==1
+replace zd_name = "Computer Science"      if zd2 ==2
+replace zd_name = "Other Egineering"      if zd2 ==3
+replace zd_name = "Other Science"         if zd2 ==4
+replace zd_name = "Semiconductors"        if zd2 ==5
+
+sort bea_code
+merge m:1 bea_code using $main/data/bea_codes
+drop if _merge ==2
+drop _merge
+
+
+* Density Distribution
+summ density14 if year ==2007 & zd_name == "Biology and Chemestry" , detail
+summ density14 if year ==2007 & zd_name == "Computer Science" , detail
+summ density14 if year ==2007 & zd_name == "Other Egineering" , detail
+summ density14 if year ==2007 & zd_name == "Other Science" , detail
+summ density14 if year ==2007 & zd_name == "Semiconductors" , detail
+
+
+sort bea_code zd2 year
+save $main/tables2/data, replace
+
+
+* TOP 10
+g top10 =0
+gsort year zd2 -density14
+by year zd2: replace top10 = 1 if _n >=1 & _n <=10
+egen top10_sum = sum(density14) if top10 ==1, by(year zd_name)
+
+summ top10_sum if year ==2007 & top10 ==1 & zd_name == "Semiconductors"
+summ top10_sum if year ==2007 & top10 ==1 & zd_name == "Computer Science"
+summ top10_sum if year ==2007 & top10 ==1 & zd_name == "Biology and Chemestry"
+
+
+sort year 
+collapse top10_sum, by(year zd_name)
+keep if year ==1971 | year ==1980 | year ==1990 | year ==2000 | year ==2007
+
+sort year
+line top10_sum  year if zd_name == "Semiconductors", legend(off) ytitle("") xtitle("")
+graph export fig7.pdf, replace
+! mv -f fig7.pdf tables2/
+
+line top10_sum  year if zd_name == "Computer Science", legend(off) ytitle("") xtitle("")
+graph export fig8.pdf, replace
+! mv -f fig8.pdf tables2/
+
+line top10_sum  year if zd_name == "Biology and Chemestry", legend(off) ytitle("") xtitle("")
+graph export fig9.pdf, replace
+! mv -f fig9.pdf tables2/
+
+
+
+
+
+************************************
+***********************************
+* PRODUCTIVITY
+************************************
+***********************************
+clear
+u $main/data2/data_3
+egen total = sum(number), by(inventor)
+
+* INCLUDE THE FOLLOWING 5 LINES IF YOU DON'T WANT INTERPOLATED DATA
+replace bea    =. if inter_bea   ==1
+replace zd     =. if inter_zd    ==1
+replace class  =. if inter_class ==1
+replace bea    =. if inter_bea2  ==1
+replace zd     =. if inter_zd2   ==1
+replace class  =. if inter_class2==1
+
+
+********************************************************************
+********************************************************************
+********************************************************************
+********************************************************************
+********************************************************************
+********************************************************************
+
+************************
+************************
+* DEFINE  Y 
+***********************
+***********************
+* NUMBER OF PATENTS IN A YEAR
+g y = log(number)
+
+
+
+***********************
+***********************
+* DEFINE   X
+***********************
+***********************
+* NUMBER OF SCIENTISTS
+*g x    = Den_bea_zd
+*g x2   = Den_bea_class
+g x   = log(Den_bea_zd    )
+*g x2  = log(Den_bea_class )
+
+* NUMBER OF FIRMS
+*g x  = Den_bea_zdB 
+*g x2 = Den_bea_classB
+
+
+***********************
+***********************
+***********************
+***********************
+keep if total  >=3.25
+
+
+
+********************************************************************
+********************************************************************
+* SUMMARY STATS FOR SAMPLE
+********************************************************************
+********************************************************************
+egen cluster1               = group(bea zd)
+keep if number~=. & zd2 ~=. & class ~=. & year ~=. & inventor ~=. & cluster1 ~=.
+
+g       corporate = 0 if org_type ~=""    & org_type ~="OT"
+replace corporate = 1 if org_type =="FI" 
+
+summ y  Den_bea_zd number total corporate
+
+* FIELD
+tab zd2
+
+
+* FIRM TYPE
+* FI      Firm
+*UN      University
+*NL      National Lab
+*RI      Research Institute
+*UG      US Government
+*HO      Hospital
+*AS      Academy of Science
+*NO      No organization recorded
+*SC      Schools (less than university level)
+*OT      Other
+tab org_type
+
+
+*******************************************************************
+********************************************************************
+*******************************
+* BINSCATTERS
+*******************************
+*******************************************************************
+********************************************************************
+ssc install binscatter
+
+binscatter y x, nquantiles(40) controls(i.bea i.year i.zd)  reportreg
+graph export figure4.pdf, replace
+! mv -f figure4.pdf tables2/
+
+
+
+
+*******************************************************************
+********************************************************************
+*******************************
+* COLLAPSE AND MERGE WITH DENSITY DATA THAT HAVE EVERY 
+* COMBINATION OF CITY-ZD-YEAR
+*
+* DATA HAVE ALL COMBINATION OF CITY-ZD-YEAR
+* I ADD 0's. THIS WAY I DO NOT DROP CELLS WITH NO PATENTS
+*
+*******************************
+*******************************
+*******************************************************************
+********************************************************************
+* COLLAPSE
+sort year bea_code zd2
+collapse y x numbe , by(year bea_co zd2)
+summ
+save $main/tmp, replace
+
+* Merge with all combinations of bea_code zd2 year
+clear
+u $main/tables2/data
+sort bea_code zd2 year
+merge 1:1  bea_code zd2 year using $main/tmp
+drop _merge
+! rm $main/tmp.dta
+rename density14 Den_bea_zd
+rename density14B Den_bea_zdB
+
+
+* I add the 0's
+replace number    = 0             if number    ==.
+
+
+*********************************
+* RANKINGS OF SIZE
+*********************************
+gsort year -Den_bea_zd
+outfile  bea_name Den_bea_zd  if year ==2007 & zd_name == "Semiconductors" using tables2/table2b.txt, replace
+outsheet bea_name Den_bea_zd  if year ==2007 & zd_name == "Semiconductors" using tables2/table2b.csv, replace comma
+
+outfile  bea_name Den_bea_zd  if year ==2007 & zd_name == "Biology and Chemestry" using tables2/table2c.txt, replace
+outsheet bea_name Den_bea_zd  if year ==2007 & zd_name == "Biology and Chemestry" using tables2/table2c.csv, replace comma
+
+outfile  bea_name Den_bea_zd  if year ==2007 & zd_name == "Computer Science" using tables2/table2d.txt, replace
+outsheet bea_name Den_bea_zd  if year ==2007 & zd_name == "Computer Science" using tables2/table2d.csv, replace comma
+
+
+
+*******************************************************************
+********************************************************************
+*******************************
+* COUNTERFACTUALS 
+*******************************
+*******************************************************************
+********************************************************************
+drop _fillin Den_bea_zdB 
+
+* Estimated elasticity
+g alpha = 0.067
+
+* PERCENTILES
+egen tmp95 = pctile(Den_bea_zd), p(95) by(year zd2)
+egen tmp90 = pctile(Den_bea_zd), p(90) by(year zd2)
+egen tmp75 = pctile(Den_bea_zd), p(75) by(year zd2)
+egen tmp50 = pctile(Den_bea_zd), p(50) by(year zd2)
+egen tmp25 = pctile(Den_bea_zd), p(25) by(year zd2)
+
+
+* This is the counterfactual cluster size. 
+* I need to do the mean, so that the total 
+* number of scientsis in the US in the 
+* counterfactual is = to observed number
+* I can't do median because it would change it
+egen mean = mean(Den_bea_zd), by(year zd2)
+
+
+* A log log model implies that the agglomeration effect in absolute amount of patents is cluster size to alpha
+* DD is estimated change in inventor productivity, measured in number of patents (not percent)
+g DD =   (mean^alpha) - (Den_bea_zd^alpha)
+* This is the percent change 
+g DDD = (mean^alpha - Den_bea_zd^alpha)*100/(Den_bea_zd^alpha)
+
+* XXX TABLE 11
+summ  DDD     if Den_bea_zd <=tmp25                       & Den_bea_zd ~=.
+summ  DDD     if Den_bea_zd >tmp25  & Den_bea_zd <= tmp50 & Den_bea_zd ~=.
+summ  DDD     if Den_bea_zd >tmp50  & Den_bea_zd <= tmp75 & Den_bea_zd ~=.
+summ  DDD     if Den_bea_zd >tmp75  & Den_bea_zd <= tmp90 & Den_bea_zd ~=.
+summ  DDD     if Den_bea_zd >tmp90  & Den_bea_zd <= tmp95 & Den_bea_zd ~=.
+summ  DDD     if Den_bea_zd >tmp95                        & Den_bea_zd ~=.
+drop tmp* 
+
+
+gsort year -Den_bea_zd
+outfile  bea_name DDD     if year ==2007 & zd_name == "Semiconductors"        using tables2/table3b.txt, replace
+outsheet bea_name DDD     if year ==2007 & zd_name == "Semiconductors"        using tables2/table3b.csv, replace comma
+outfile  bea_name DDD     if year ==2007 & zd_name == "Biology and Chemestry" using tables2/table3c.txt, replace
+outsheet bea_name DDD     if year ==2007 & zd_name == "Biology and Chemestry" using tables2/table3c.csv, replace comma
+outfile  bea_name DDD     if year ==2007 & zd_name == "Computer Science"      using tables2/table3d.txt, replace
+outsheet bea_name DDD     if year ==2007 & zd_name == "Computer Science"      using tables2/table3d.csv, replace comma
+
+
+* Change
+g ZZ =  Den_bea_zd*DD
+g ZZZ = Den_bea_zd*(Den_bea_zd^alpha)
+preserve
+collapse (sum)  ZZ ZZZ, by(year zd_name)
+
+* Aggregate effect by field
+* XXX TABLE 12 col 1
+g change = ZZ / ZZZ
+list zd change if year ==2007
+
+
+* Aggregate effect -- All fields
+* XXX TABLE 12  col 1
+restore
+collapse (sum) ZZ ZZZ, by(year)
+g change = ZZ / ZZZ
+list change if year ==2007
+
+

107/replication_package/AER_UPLOADED/reg20.do

@@ -0,0 +1,605 @@
+********************************************************************
+********************************************************************
+********************************************************************
+* I ESTIMATE A NUMBER OF EXTENSIONS AND ROBUSNESS CHECKS
+* I ALSO COMPUTE  NEW SUMMARY STATS 
+********************************************************************
+********************************************************************
+********************************************************************
+
+clear
+u $main/data2/data_3
+
+egen total = sum(number), by(inventor)
+
+* INCLUDE THE FOLLOWING 6 LINES IF YOU DON'T WANT INTERPOLATED DATA
+replace bea    =. if inter_bea   ==1
+replace zd     =. if inter_zd    ==1
+replace class  =. if inter_class ==1
+replace bea    =. if inter_bea2  ==1
+replace zd     =. if inter_zd2   ==1
+replace class  =. if inter_class2==1
+
+
+
+* CLUSTER SIZE 
+g x   = log(Den_bea_zd    )
+sort inventor year
+by inventor : g  x_m1 = x[_n-1]
+by inventor : g  x_m2 = x[_n-2]
+by inventor : g  x_m3 = x[_n-3]
+by inventor : g  x_m4 = x[_n-4]
+by inventor : g  x_m5 = x[_n-5]
+by inventor : g  x_p1 = x[_n+1]
+by inventor : g  x_p2 = x[_n+2]
+by inventor : g  x_p3 = x[_n+3]
+by inventor : g  x_p4 = x[_n+4]
+by inventor : g  x_p5 = x[_n+5]
+
+
+drop if year  ==.
+drop if zd ==.
+drop if bea ==.
+egen org_new2 = group(org_id) if org_id ~= ""
+
+
+
+********************************************************************
+********************************************************************
+********************************************************************
+********************************************************************
+********************************************************************
+********************************************************************
+
+* NUMBER OF PATENTS IN A YEAR
+g y = log(number)
+
+* CITATIONS
+* To avoid dropping the 0's, I add 0.00001
+g y2 = log(citations+0.00001)
+g y3 = log( (citations+0.00001) / number) 
+
+* SPECIALIZATION
+*g y4 = 1-general
+
+* NUMBER OF FIRMS
+g x3  = log(Den_bea_zdB) 
+
+
+***********************
+***********************
+* STAR INVENTORS
+* top 10% = 3.25
+***********************
+***********************
+keep if total  >=3.25
+
+
+
+********************************************************************
+********************************************************************
+********************************************************************
+* NEW VARIABLES
+********************************************************************
+********************************************************************
+********************************************************************
+egen cluster1               = group(bea zd)
+egen cluster0               = group(bea zd year)
+egen cluster2               = group(bea )
+egen cluster_bea_year       = group(bea year)
+egen cluster_bea_zd_year    = group(bea zd year)
+egen cluster_bea_class_year = group(bea class year)
+egen cluster_zd_year        = group(zd year)
+egen cluster_bea_class      = group(bea class)
+egen cluster_class_year     = group(class year)
+egen org_new                = group(org_id)
+egen cluster_org_year       = group(org_new year)
+egen org_size               = count(inventor), by(org_new year)
+egen lab_size              = count(inventor), by(org_new year bea)
+
+
+keep if y~=. & zd ~=. & class ~=. & year ~=. & inventor ~=. & cluster1 ~=. & bea ~=.
+summ y y2 y3  x
+summ number, detail
+summ x Den_bea_zd , detail
+tab zd
+
+
+save $main/data2/tmp10, replace
+
+********************************************************************
+********************************************************************
+********************************************************************
+* NON LINEARITIES
+********************************************************************
+********************************************************************
+********************************************************************
+g x_sq = x*x
+g x_cu = x*x*x
+reghdfe y x x_sq x_cu, absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor cluster_bea_year org_new ) vce(cluster cluster1) keepsin
+
+test x_sq = x_cu = 0
+
+
+
+********************************************************************
+********************************************************************
+********************************************************************
+* CROSS-FIELD SPILLOVERS 
+********************************************************************
+********************************************************************
+********************************************************************
+sort bea year
+merge m:1 bea year using $main/data2/density14_3_reshaped
+drop _merge
+
+
+
+* Average cluster size in other fields in the same city
+******************
+
+egen x_other_fields1 = rowmean(density2 density3 density4 density5) if zd ==1
+egen x_other_fields2 = rowmean(density1 density3 density4 density5) if zd ==2
+egen x_other_fields3 = rowmean(density1 density2 density4 density5) if zd ==3
+egen x_other_fields4 = rowmean(density1 density2 density3 density5) if zd ==4
+egen x_other_fields5 = rowmean(density1 density2 density3 density4) if zd ==5
+g       x_other_fields = x_other_fields1 if zd ==1
+replace x_other_fields = x_other_fields2 if zd ==2
+replace x_other_fields = x_other_fields3 if zd ==3
+replace x_other_fields = x_other_fields4 if zd ==4
+replace x_other_fields = x_other_fields5 if zd ==5
+
+* Logs
+replace x_other_fields = log(x_other_fields)
+replace density1 = log(density1)
+replace density2 = log(density2)
+replace density3 = log(density3)
+replace density4 = log(density4)
+replace density5 = log(density5)
+
+* Correlations 
+reg x_other_fields x
+
+reghdfe x_other_fields x, absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor cluster_bea_year org_new                                   ) vce(cluster cluster1) keepsin
+
+* Cross-field spillovers -- All fields together
+eststo: reghdfe y x x_other_fields, absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor cluster_bea_year org_new                                   ) vce(cluster cluster1) keepsin
+
+* Cross-field spillovers --  5 X 5 matrix
+* Note: I need to drop bea*year effects due to multicollinearity
+
+eststo: reghdfe y density1 density2 density3 density4 density5 if zd ==1, absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor org_new                                   ) vce(cluster cluster1) keepsin
+
+eststo: reghdfe y density1 density2 density3 density4  density5 if zd ==2, absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor org_new                                   ) vce(cluster cluster1) keepsin
+
+eststo: reghdfe y density1 density2 density3 density4  density5 if zd ==3, absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor org_new                                   ) vce(cluster cluster1) keepsin
+
+eststo: reghdfe y density1 density2 density3 density4 density5 if zd ==4, absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor org_new                                   ) vce(cluster cluster1) keepsin
+
+eststo: reghdfe y density1 density2 density3 density4  density5 if zd==5, absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor org_new                                   ) vce(cluster cluster1) keepsin
+
+esttab using tables20/table3.tex, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+esttab using tables20/table3, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+eststo clear
+
+
+
+*****************************************
+**********************************************
+* MEASURES OF CLUSTER SIZE THAT PUT MORE WEIGHT ON BETTER INVENTORS 
+**********************************************
+**********************************************
+sort  inventor year 
+merge m:1 inventor year using $main/data2/density14_W
+tab _merge
+drop _merge 
+
+
+* WEIGHT INVENTORS BASED ON NUMBER OF PATENTS OR CITATIONS
+g x6   = log(Den_bea_zd_WW4)
+
+eststo: reghdfe y x6, absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor cluster_bea_year org_new                                   ) vce(cluster cluster1) keepsin
+
+* COUNT INVENTORS WITH NUMBER OF PATENTS ABOVE A CERTAIN THRESHOLD
+drop x6
+g x6   = log(Den_bea_zd_WW1 )
+
+eststo: reghdfe y x6, absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor cluster_bea_year org_new                                   ) vce(cluster cluster1) keepsin
+
+
+* WEIGHT INVENTORS BASED ON NUMBER OF PATENTS 
+drop x6
+g x6   = log(Den_bea_zd_W4)
+
+eststo: reghdfe y x6, absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor cluster_bea_year org_new                                   ) vce(cluster cluster1) keepsin
+
+* COUNT INVENTORS WITH NUMBER OF CITATIONS ABOVE A CERTAIN THRESHOLD
+drop x6
+g x6   = log(Den_bea_zd_W1 )
+
+eststo: reghdfe y x6, absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor cluster_bea_year org_new                                   ) vce(cluster cluster1) keepsin
+
+
+
+********************************************************************
+********************************************************************
+********************************************************************
+* TEAMS 
+********************************************************************
+********************************************************************
+********************************************************************
+summ team team2, detail
+
+* Control for quadratic in team size
+drop x6
+g x6 = x
+g team_sq = team*team
+eststo: reghdfe y x6 team team_sq, absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor cluster_bea_year org_new                                   ) vce(cluster cluster1) keepsin
+
+
+* Density is measured excluding all members of the team
+sort  inventor year
+merge m:1 inventor_id year using $main/data2/density_team
+drop _merge
+drop x6
+g x6   = log(density14_team)
+
+eststo: reghdfe y x6, absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor cluster_bea_year org_new                                   ) vce(cluster cluster1) keepsin
+
+
+
+* Interact x with indicator for large team size
+g cc = (team >=2)
+g x_cc = x*cc
+eststo: reghdfe y x6 x_cc cc , absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor cluster_bea_year org_new                                   ) vce(cluster cluster1) keepsin
+
+
+* Solo inventors  -- Interactions
+* Team2 is the share of patents that an inventor created alone in a given year
+g jj      = (team2 >=.9)
+g x_team2 = x*(team2 >=.9)
+drop x6
+g x6 =x
+eststo: reghdfe y x6 x_team2 jj, absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor cluster_bea_year org_new                                   ) vce(cluster cluster1) keepsin
+
+
+esttab using tables20/table87.tex, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+esttab using tables20/table87, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+eststo clear
+
+
+********************************************************************
+********************************************************************
+********************************************************************
+* ADDITIONAL SPECIFICATIONS
+********************************************************************
+********************************************************************
+********************************************************************
+
+**********************************************
+* Density defined at city level 
+* std errors are clustered by city
+* I need to drop city*year effects
+g xx   = log(Den_bea    )
+eststo: reghdfe y xx , absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor org_new) vce(cluster cluster2) keepsin
+
+
+**********************************************
+* std errors are clustered by city in main model 
+eststo: reghdfe y x , absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor cluster_bea_year org_new                                   ) vce(cluster cluster2) keepsin
+
+
+
+*************************************************************
+* Lifecycle controls 
+sort inventor year
+by inventor: g ttt = _n
+g first_year = year if ttt ==1
+egen first_year2 = min(first_year), by(inventor)
+g age = year - first_year2
+summ year ttt first*
+drop ttt first_*
+eststo: reghdfe y x , absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor cluster_bea_year org_new  age                                 ) vce(cluster cluster1) keepsin
+
+
+esttab using tables20/table7.tex, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+esttab using tables20/table7, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+eststo clear
+
+
+
+********************************************************
+****************************************************
+************************************************
+* Field Specific coefficients (REF 1)
+* need to drop city*year effects
+************************************************
+****************************************************
+*******************************************************
+* Comp
+eststo: reghdfe y x if zd ==2, absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor org_new                                   ) vce(cluster cluster1) keepsin
+* Bio
+eststo: reghdfe y x if zd ==1, absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor org_new                                   ) vce(cluster cluster1) keepsin
+* Semi
+eststo: reghdfe y x if zd ==5, absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor org_new                                   ) vce(cluster cluster1) keepsin
+* Other eng
+eststo: reghdfe y x if zd ==3, absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor org_new                                   ) vce(cluster cluster1) keepsin
+* Other Sci
+eststo: reghdfe y x if zd ==4, absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor org_new                                   ) vce(cluster cluster1) keepsin
+
+esttab using tables20/table8.tex, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+esttab using tables20/table8, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+eststo clear
+
+
+*************************************************************
+*************************************************************
+*************************************************************
+* Heterogeneity of the effect by firm productivity 
+*************************************************************
+*************************************************************
+*************************************************************
+egen firm_output = mean(y), by(org_new zd)
+
+summ firm_output , detail
+egen pctile25 = pctile(firm_output) if y~=., p(25) by(zd )
+egen pctile50 = pctile(firm_output) if y~=., p(50) by(zd )
+egen pctile75 = pctile(firm_output) if y~=., p(75) by(zd )
+g dd1 = 0 if firm_output ~=.
+g dd2 = 0 if firm_output ~=.
+g dd3 = 0 if firm_output ~=.
+g dd4 = 0 if firm_output ~=.
+replace dd1 =1 if firm_output                            <= pctile25 & firm_output ~=.
+replace dd2 =1 if firm_output  > pctile25 & firm_output  <= pctile50 & firm_output ~=.
+replace dd3 =1 if firm_output  > pctile50 & firm_output  <= pctile75 & firm_output ~=.
+replace dd4 =1 if firm_output  > pctile75                            & firm_output ~=.
+summ dd1 dd2 dd3 dd4
+
+g x_dd1 = x*dd1
+g x_dd2 = x*dd2
+g x_dd3 = x*dd3
+g x_dd4 = x*dd4
+
+
+
+eststo: reghdfe y x_dd1 x_dd2 x_dd3 x_dd4 dd1 dd2 dd3 dd4, absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor cluster_bea_year ) vce(cluster cluster1) keepsin
+
+test x_dd1 =x_dd2 =x_dd3 =x_dd4
+
+eststo: reghdfe y x_dd1 x_dd2 x_dd3 x_dd4 dd1 dd2 dd3 dd4, absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor cluster_bea_year org_new) vce(cluster cluster1) keepsin
+
+test x_dd1 =x_dd2 =x_dd3 =x_dd4
+
+
+esttab using tables20/table2.tex, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+esttab using tables20/table2, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+eststo clear
+
+
+
+*************************************************************
+*************************************************************
+*************************************************************
+* QUALITY: LOCAL CITATIONS RECEIVED 
+*************************************************************
+*************************************************************
+*************************************************************
+clear
+u $main/data2/tmp10
+merge 1:1 inventor year using $main/data2/citations_received
+drop _merge
+
+* Number of citations received by an inventor in a given year that are from same bea
+summ same_bea
+g y2_local = log(  (citations*same_bea)+0.00001)
+g y3_local = log(( (citations*same_bea) +0.00001) / number)
+* Number of citations received by an inventor in a given year that are from different bea
+g y2_nolocal = log(  (citations*(1-same_bea))+0.00001)
+g y3_nolocal = log(( (citations*(1-same_bea)) +0.00001) / number)
+
+
+* Local 
+g YY = y2_local 
+eststo: reghdfe YY x , absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor cluster_bea_year org_new  ) vce(cluster cluster1) keepsin
+
+* No Local
+drop YY
+g YY = y2_nolocal 
+eststo: reghdfe YY x , absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor cluster_bea_year org_new  ) vce(cluster cluster1) keepsin
+
+esttab using tables20/table6.tex, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+esttab using tables20/table6, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+eststo clear
+
+* Local
+drop YY
+g YY = y3_local 
+eststo: reghdfe YY x , absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor cluster_bea_year org_new  ) vce(cluster cluster1) keepsin
+
+* No local
+drop YY
+g YY = y3_nolocal
+eststo: reghdfe YY x , absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor cluster_bea_year org_new  ) vce(cluster cluster1) keepsin
+
+esttab using tables20/table7.tex, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+esttab using tables20/table7, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+eststo clear
+
+
+
+*************************************************************
+*************************************************************
+*************************************************************
+* QUALITY: GENERALITY AND ORIGINALITY 
+*************************************************************
+*************************************************************
+*************************************************************
+summ general original
+drop YY
+g YY = general
+eststo: reghdfe YY x , absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor cluster_bea_year org_new                                   ) vce(cluster cluster1) keepsin
+
+drop YY
+g YY = original  
+eststo: reghdfe YY x , absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor cluster_bea_year org_new                                   ) vce(cluster cluster1) keepsin
+
+
+eststo clear
+
+
+
+*************************************************************
+*************************************************************
+*************************************************************
+* LOCAL CITATIONS MADE 
+*************************************************************
+*************************************************************
+*************************************************************
+* identify stars
+clear
+u $main/data2/tmp10
+g star =1
+collapse star, by(inventor_id)
+sort inventor_id
+save $main/data2/stars, replace
+
+
+clear
+u $main/data2/citations_made
+sort inventor_id
+merge m:1 inventor_id using $main/data2/stars
+
+
+* Descriptive means:
+g diff_zd = 1-same_zd 
+summ same_bea same_zd diff_zd same_bea_zd  same_bea_diff_zd if star ==1
+summ same_bea same_zd diff_zd same_bea_zd  same_bea_diff_zd if star ~=1
+clear
+
+
+* Effect of cluster size on citations, local citations, share of local citations
+u $main/data2/tmp10
+sort inventor_id year
+merge 1:1 inventor_id year using $main/data2/citations_made
+
+keep if x ~=.
+summ x, detail
+g N_same_bea = log(citations_made*same_bea )
+g log_citations_made = log(citations_made)
+
+g N_same_bea2 = log(citations_made*same_bea + 0.001)
+g log_citations_made2 = log(citations_made + 0.001)
+
+
+summ citations_made
+g vv = citations_made*same_bea 
+summ vv
+g vvv = citations_made*same_zd
+summ vvv
+
+
+
+summ log_citations_made 
+summ log_citations_made2 
+
+summ N_same_bea 
+summ N_same_bea2 
+
+summ same_bea 
+summ same_zd
+
+
+g YY = log_citations_made 
+eststo: reghdfe YY x, absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor cluster_bea_year org_new                                   ) vce(cluster cluster1) keepsin
+
+drop YY
+g YY = log(citations_made/number)
+eststo: reghdfe YY x, absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor cluster_bea_year org_new                                   ) vce(cluster cluster1) keepsin
+
+
+drop YY
+g YY =same_bea
+eststo: reghdfe YY x, absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor cluster_bea_year org_new                                   ) vce(cluster cluster1) keepsin
+
+
+drop YY
+g YY = same_zd
+eststo: reghdfe YY x, absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor cluster_bea_year org_new                                   ) vce(cluster cluster1) keepsin
+
+
+
+esttab using tables20/table72.tex, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+esttab using tables20/table72, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+eststo clear
+
+
+**********************************************
+* I estimate baseline model on inventors who tend to cite patents in same field
+* same_zd is a share, not a dummy
+summ same_zd, detail 
+g bb = (same_zd>=.9)
+g x_bb = x*bb
+eststo: reghdfe y x x_bb bb , absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor cluster_bea_year org_new                                   ) vce(cluster cluster1) keepsin
+
+
+
+
+
+***************************************
+* SUMMARY STATISTICS 
+****************************************
+clear
+* patent level data
+u $main/data2/data_patent_level_3
+sort inventor_id
+merge m:1 inventor_id using $main/data2/stars
+tab _merge
+drop _merge
+replace star = 0 if star ==.
+
+* tag one obs in each patent-city combination
+sort patent_id bea
+by patent_id bea: g hhh = _n
+g h2 = 1 if hhh ==1
+
+preserve
+collapse star citations general original (count) team_size = inventor_id (sum) n_star = star n_bea = h2, by(patent_id)
+summ
+
+* team size
+summ team_size if n_star >=1
+summ team_size if n_star ==1
+summ team_size if n_star ==0
+
+* share of stars in a team
+sum star
+* number of stars in a team
+sum n_star
+
+* number of cities in the team
+summ n_bea if n_star >=1
+summ n_bea if n_star ==0
+
+* number and type of citations
+summ citations if n_star >=1
+summ citations if n_star ==0
+summ general if n_star >=1
+summ general if n_star ==0
+summ original if n_star >=1
+summ original if n_star ==0
+
+
+********************************
+* number of fields an by inventor
+restore
+sort inventor_id zd2
+by inventor_id zd2: g first = _n
+keep if first ==1
+
+collapse  (count) zd2, by(inventor_id star)
+summ zd2 if star ==1
+summ zd2 if star ==0
+
+
+
+

107/replication_package/AER_UPLOADED/reg22.do

@@ -0,0 +1,330 @@
+
+********************************************************************
+********************************************************************
+********************************************************************
+* KODAK -- MATCHING  
+********************************************************************
+********************************************************************
+********************************************************************
+
+
+u $main/data2/data_3 
+
+replace bea    =. if inter_bea   ==1
+replace zd     =. if inter_zd    ==1
+replace class  =. if inter_class ==1
+replace bea    =. if inter_bea2  ==1
+replace zd     =. if inter_zd2   ==1
+replace class  =. if inter_class2==1
+
+
+g y = log(number)
+g x   = log(Den_bea_zd    )
+
+egen cluster1               = group(bea zd)
+egen cluster_bea_year       = group(bea year)
+egen cluster_bea_zd_year    = group(bea zd year)
+egen cluster_class_year = group(bea class year)
+egen cluster_zd_year        = group(zd year)
+egen cluster_bea_class      = group(bea class)
+
+
+
+******************************************************************************
+* SAMPLE
+* I KEEP ONLY NON-KODAK INVENTORS
+* I KEEP ONLY 96 and 07
+******************************************************************************
+g kodak = 1 if org_id =="FI_1757"
+g xerox = 1 if org_id =="FI_6225"
+
+keep if y~=. & zd ~=. & class ~=. & year ~=. & inventor ~=. & cluster1 ~=. & bea ~=.
+*keep if kodak ~=1
+drop if class ==396 | class ==399 
+keep if year ==1996 | year ==2007
+
+keep if kodak ~=1
+
+
+* DUMMIES
+****************
+* rochester dummy
+g roch = 0
+replace roch = 1 if bea ==139
+
+* dummy for 2007
+g       after = 0 if year ==1996
+replace after = 1 if year ==2007
+
+* Rochester * 2007
+g roch_after = roch*after
+
+
+
+
+* Compute population distribution in bea-level data
+preserve
+clear
+u $main/data/read
+sum
+g roch = 0
+replace roch = 1 if bea ==139
+summ pop if roch ==1
+summ pop if roch ~=1, detail
+clear
+
+
+* Add Population and other covariates
+restore
+sort bea
+merge m:1 bea using $main/data/read
+drop _merge
+
+
+* Mean productvity and field distribution in 1996, by city
+preserve
+keep if year ==1996
+g y_bea = y
+g D1 = (zd2==1)
+g D2 = (zd2==2)
+g D3 = (zd2==3)
+g D4 = (zd2==4)
+g D5 = (zd2==5)
+
+collapse y_bea D1 D2 D3 D4 D5, by(bea_code)
+
+
+summ y_bea if bea ==139, detail
+summ y_bea if bea ~=139, detail
+sort bea
+save tables22/bea_data, replace
+clear
+
+
+
+**************************************************************
+**************************************************************
+**************************************************************
+**************************************************************
+* VERSION 1:  WEIGHT
+**************************************************************
+**************************************************************
+**************************************************************
+**************************************************************
+restore
+merge m:1 bea_code using $main/tables22/bea_data
+drop  _merge
+save tmp22, replace
+
+
+summ D1-D5 if roch ==1 & year ==1996
+
+* D1 D3 and D4 are the 3 largest fields in Roshester in 1996
+logit roch medfaminc90 trade90 agr90  const9 manuf90 nonwhite90 pop emp D1 D3 D4 y_bea if year == 1996
+
+predict  predicted
+summ year predicted
+g WW = 1/predicted if roch ==1
+replace WW = 1/(1-predicted) if roch ==0
+
+
+
+*****************
+* CROSS-SECIONAL
+*****************
+g cluster3 = bea
+preserve
+eststo:      reg y roch_after after roch [weight = WW],                                                      cluster(cluster3)
+eststo: reghdfe  y roch_after after roch [weight = WW], absorb(zd)                                           cluster(cluster3) keepsin
+eststo: reghdfe  y roch_after after roch [weight = WW], absorb(cluster_zd_year)                              cluster(cluster3) keepsin
+eststo: reghdfe  y roch_after after roch [weight = WW], absorb(cluster1 cluster_zd_year)                     cluster(cluster3) keepsin
+
+esttab using tables22/table2.tex, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+esttab using tables22/table2, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+eststo clear
+
+
+
+
+***************************
+* 2SLS
+***************************
+* First stage Effect on cluster size
+eststo: reg x roch_after after roch [weight = WW],                                                      cluster(cluster3)
+test roch_after ==0
+eststo: reghdfe  x roch_after after roch [weight = WW], absorb(zd)                                           cluster(cluster3) keepsin
+test roch_after ==0
+eststo: reghdfe  x roch_after after roch [weight = WW], absorb(cluster_zd_year)                              cluster(cluster3) keepsin
+test roch_after ==0
+eststo: reghdfe  x roch_after after roch [weight = WW], absorb(cluster1 cluster_zd_year)                     cluster(cluster3) keepsin
+test roch_after ==0
+esttab using tables22/table5.tex, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+esttab using tables22/table5, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+eststo clear
+
+* IV
+eststo: ivreg     y (x =roch_after) after roch [weight = WW],                                  cluster(cluster3)
+eststo: ivreghdfe y (x =roch_after) after roch [weight = WW], absorb(zd                      ) cluster(cluster3)
+eststo: ivreghdfe y (x =roch_after) after roch [weight = WW], absorb(cluster_zd_year         ) cluster(cluster3)
+eststo: ivreghdfe y (x =roch_after) after roch [weight = WW], absorb(cluster1 cluster_zd_year) cluster(cluster3)
+
+
+esttab using tables22/table6.tex, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+esttab using tables22/table6, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+eststo clear
+
+
+
+
+****************************************
+* WITHIN INVENTOR DIFFERENCES
+*****************************************
+restore
+egen zzz = count(y), by(inventor)
+tab zzz
+keep if zzz ==2
+drop zzz
+preserve
+
+
+* New rochester dummy for rochester status in 1996
+* Note: if in Rochester in 1996, this dummy stays =1 both in 1996 and  2007
+drop roch roch_after
+g       tmp6 = 0
+replace tmp6 = 1 if bea ==139 & year == 1996
+egen roch = max(tmp6), by(inventor)
+g roch_after = roch*after
+
+eststo: reghdfe  y roch_after after roch [weight = WW], absorb(inventor)                           cluster(cluster3) keepsin
+eststo: reghdfe  y roch_after after roch [weight = WW], absorb(inventor zd)                        cluster(cluster3) keepsin
+eststo: reghdfe  y roch_after after roch [weight = WW], absorb(inventor cluster_zd_year)           cluster(cluster3) keepsin
+eststo: reghdfe  y roch_after after roch [weight = WW], absorb(inventor cluster_zd_year cluster1 ) cluster(cluster3) keepsin
+
+esttab using tables22/table3.tex, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+esttab using tables22/table3, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+eststo clear
+
+
+
+* New rochester dummy for rochester status in 1996 and 2007
+* Note: it is =1 only of inventor is in rochester both in 1996 and  2007
+clear
+restore
+drop roch roch_after
+
+g tmp6 = 1 if bea ==139 & year == 1996
+g tmp7 = 1 if bea ==139 & year == 2007
+egen tmp8 = max(tmp6), by(inventor)
+egen tmp9 = max(tmp7), by(inventor)
+g roch = 0
+replace roch = 1 if tmp8 ==1 & tmp9 ==1
+g roch_after = roch*after
+drop tmp*
+
+eststo: reghdfe  y roch_after after roch [weight = WW], absorb(inventor)                           cluster(cluster3) keepsin
+eststo: reghdfe  y roch_after after roch [weight = WW], absorb(inventor zd)                        cluster(cluster3) keepsin
+eststo: reghdfe  y roch_after after roch [weight = WW], absorb(inventor cluster_zd_year)           cluster(cluster3) keepsin
+eststo: reghdfe  y roch_after after roch [weight = WW], absorb(inventor cluster_zd_year cluster1 ) cluster(cluster3) keepsin
+
+esttab using tables22/table4.tex, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+esttab using tables22/table4, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+eststo clear
+
+
+**************************************************************
+**************************************************************
+**************************************************************
+**************************************************************
+* VERSION 2: I DROP CITIES THAT ARE DIFFERENT FROM ROCHESTER
+* BASED ON INITIAL CHARACERISTICS
+**************************************************************
+**************************************************************
+**************************************************************
+**************************************************************
+clear
+u tmp22
+keep if pop  >=580511                   
+keep if medfaminc90 >= 27397
+keep if y_bea >= -.77 & y_bea <= -.57
+keep if D3 >= .40
+
+
+
+
+*****************
+* CROSS-SECIONAL
+*****************
+g cluster3 = bea
+eststo:      reg y roch_after after roch,                                                      cluster(cluster3)  
+eststo: reghdfe  y roch_after after roch, absorb(zd)                                           cluster(cluster3) keepsin 
+eststo: reghdfe  y roch_after after roch, absorb(cluster_zd_year)                              cluster(cluster3) keepsin 
+eststo: reghdfe  y roch_after after roch, absorb(cluster1 cluster_zd_year)                     cluster(cluster3) keepsin 
+
+esttab using tables22/table2b.tex, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+esttab using tables22/table2b, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+eststo clear
+
+
+
+
+****************************************
+* WITHIN INVENTOR 
+*****************************************
+egen zzz = count(y), by(inventor)
+tab zzz
+keep if zzz ==2
+drop zzz
+preserve
+
+* New rochester dummy for rochester status in 1996
+* Note: if in Rochester in 1996, this dummy stays =1 both in 1996 and  2007
+drop roch roch_after
+g       tmp6 = 0
+replace tmp6 = 1 if bea ==139 & year == 1996
+egen roch = max(tmp6), by(inventor)
+g roch_after = roch*after
+
+eststo: reghdfe  y roch_after after roch, absorb(inventor)                           cluster(cluster3) keepsin
+eststo: reghdfe  y roch_after after roch, absorb(inventor zd)                        cluster(cluster3) keepsin
+eststo: reghdfe  y roch_after after roch, absorb(inventor cluster_zd_year)           cluster(cluster3) keepsin
+eststo: reghdfe  y roch_after after roch, absorb(inventor cluster_zd_year cluster1 ) cluster(cluster3) keepsin
+
+esttab using tables22/table3b.tex, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+esttab using tables22/table3b, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+eststo clear
+
+
+* New rochester dummy for rochester status in 1996 and 2007
+* Note: it is =1 only of inventor is in rochester both in 1996 and  2007
+clear
+restore
+preserve
+drop roch roch_after
+
+g tmp6 = 1 if bea ==139 & year == 1996
+g tmp7 = 1 if bea ==139 & year == 2007
+egen tmp8 = max(tmp6), by(inventor)
+egen tmp9 = max(tmp7), by(inventor)
+g roch = 0
+replace roch = 1 if tmp8 ==1 & tmp9 ==1
+g roch_after = roch*after
+drop tmp*
+
+eststo: reghdfe  y roch_after after roch, absorb(inventor)                           cluster(cluster3) keepsin
+eststo: reghdfe  y roch_after after roch, absorb(inventor zd)                        cluster(cluster3) keepsin
+eststo: reghdfe  y roch_after after roch, absorb(inventor cluster_zd_year)           cluster(cluster3) keepsin
+eststo: reghdfe  y roch_after after roch, absorb(inventor cluster_zd_year cluster1 ) cluster(cluster3) keepsin
+
+esttab using tables22/table4b.tex, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+esttab using tables22/table4b, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+eststo clear
+
+
+
+rm tmp22.dta
+restore
+
+
+
+
+

107/replication_package/AER_UPLOADED/reg23.do

@@ -0,0 +1,304 @@
+********************************************************************
+********************************************************************
+********************************************************************
+* MOVERS - LEADS AND LAGS
+********************************************************************
+********************************************************************
+********************************************************************
+
+********************************************************************
+********************************************************************
+********************************************************************
+* READ DATA
+********************************************************************
+********************************************************************
+********************************************************************
+clear
+u $main/data2/data_3
+
+egen total = sum(number), by(inventor)
+
+* INCLUDE THE FOLLOWING 6 LINES IF YOU DON'T WANT INTERPOLATED DATA
+replace bea    =. if inter_bea   ==1
+replace zd     =. if inter_zd    ==1
+replace class  =. if inter_class ==1
+replace bea    =. if inter_bea2  ==1
+replace zd     =. if inter_zd2   ==1
+replace class  =. if inter_class2==1
+
+
+
+* CLUSTER SIZE
+g x   = log(Den_bea_zd    )
+sort inventor year
+by inventor : g  x_m1 = x[_n-1]
+by inventor : g  x_m2 = x[_n-2]
+by inventor : g  x_m3 = x[_n-3]
+by inventor : g  x_m4 = x[_n-4]
+by inventor : g  x_m5 = x[_n-5]
+by inventor : g  x_p1 = x[_n+1]
+by inventor : g  x_p2 = x[_n+2]
+by inventor : g  x_p3 = x[_n+3]
+by inventor : g  x_p4 = x[_n+4]
+by inventor : g  x_p5 = x[_n+5]
+
+
+
+drop if year  ==.
+drop if zd ==.
+drop if bea ==.
+egen org_new2 = group(org_id) if org_id ~= ""
+
+
+
+********************************************************************
+********************************************************************
+********************************************************************
+********************************************************************
+********************************************************************
+********************************************************************
+
+* NUMBER OF PATENTS IN A YEAR
+g y = log(number)
+
+* CITATIONS
+* To avoid dropping the 0's, I add 0.00001
+g y2 = log(citations+0.00001)
+g y3 = log( (citations+0.00001) / number)
+
+* SPECIALIZATION
+*g y4 = 1-general
+
+* NUMBER OF FIRMS
+g x3  = log(Den_bea_zdB)
+
+
+***********************
+***********************
+* DEFINE STAR INVENTORS
+***********************
+***********************
+keep if total  >=3.25
+
+
+
+********************************************************************
+********************************************************************
+********************************************************************
+* NEW VARIABLES
+********************************************************************
+********************************************************************
+********************************************************************
+egen cluster1               = group(bea zd)
+egen cluster0               = group(bea zd year)
+egen cluster2               = group(bea )
+egen cluster_bea_year       = group(bea year)
+egen cluster_bea_zd_year    = group(bea zd year)
+egen cluster_bea_class_year = group(bea class year)
+egen cluster_zd_year        = group(zd year)
+egen cluster_bea_class      = group(bea class)
+egen cluster_class_year     = group(class year)
+egen org_new                = group(org_id)
+egen cluster_org_year       = group(org_new year)
+egen org_size               = count(inventor), by(org_new year)
+egen lab_size              = count(inventor), by(org_new year bea)
+
+
+
+********************************************************************
+********************************************************************
+********************************************************************
+* REGRESSIONS
+********************************************************************
+********************************************************************
+********************************************************************
+keep if y~=. & zd ~=. & class ~=. & year ~=. & inventor ~=. & cluster1 ~=. & bea ~=.
+summ y y2 y3  x
+summ number, detail
+summ x Den_bea_zd , detail
+tab zd
+
+
+************************************************************************
+************************************************************************
+
+preserve
+
+* Mover dummy
+rename bea_code bea
+sort inventor year
+by inventor : g  bea_m1 = bea[_n-1]
+by inventor : g  year_m1 = year[_n-1]
+g       move = 0
+replace move = 1 if bea ~= bea_m1 & bea ~=. & bea_m1 ~=.
+
+
+
+* Indicator for cases where an inventor is in a different city  relative
+* to last year she was observed, and last year observed is more than 1 year ago
+* In these case, the exact time of the move in unknown
+g dyear = year - year_m1
+g large_gap = 1 if move ==1 & dyear >1
+* Assign this dummy to all years when an inventor is observed
+egen gapp = max(large_gap), by(inventor)
+drop large_gap dyear
+
+g move_year = year if move ==1
+* Total number of moves by inventor
+egen nn1 = total(move), by(inventor)
+
+
+* move_year1 is year of first observed move
+* It is defined only for those who move once
+egen move_year1 = min(move_year),  by(inventor)
+*drop move_year
+replace move_year1 = . if nn>1
+* Time since move
+g tt = year - move_year1
+
+* Keep only movers, when time of the move is exactly identified
+* For coding simplicity, I keep those who move once
+keep if nn1 ==1 
+drop if gapp ==1
+
+
+* Mean x in the 5 years before the move and the five years after the move
+egen tmp_mm = mean(x) if tt >= -5 & tt <=-1, by(inventor)
+egen tmp_pp = mean(x) if tt >= 1  & tt <= 5, by(inventor)
+
+* tmp_mm and tmp_pp are constants, but they are defined only in the 5 years 
+* before the move and the 5 years after the move, repsctively. To run the regression, 
+* I assign tmp_mm and tmp_pp to an entire inventor life so that they are not missing
+egen tmp_m = max(tmp_mm),  by(inventor)
+egen tmp_p = max(tmp_pp),  by(inventor)
+drop tmp_pp tmp_mm
+
+
+* Now I interact the average x in the years before and after the move with
+* indicators for numbeer of years since the move 
+g m1 = (tt==-1)
+g m2 = (tt==-2)
+g m3 = (tt==-3)
+g m4 = (tt==-4)
+g m5 = (tt==-5)
+
+g p1 = (tt==1)
+g p2 = (tt==2)
+g p3 = (tt==3)
+g p4 = (tt==4)
+g p5 = (tt==5)
+
+drop x_p* x_m*
+g  x_m1 = tmp_m*m1
+g  x_m2 = tmp_m*m2
+g  x_m3 = tmp_m*m3
+g  x_m4 = tmp_m*m4
+g  x_m5 = tmp_m*m5
+g  x_p1 = tmp_p*p1
+g  x_p2 = tmp_p*p2
+g  x_p3 = tmp_p*p3
+g  x_p4 = tmp_p*p4
+g  x_p5 = tmp_p*p5
+
+
+
+
+
+* Regression
+eststo: reghdfe y x_p5 x_p4 x_p3 x_p2 x_p1 x x_m1 x_m2 x_m3 x_m4 x_m5 ,absorb(year bea zd class cluster1 cluster_bea_class cluster_zd_year cluster_class_year inventor cluster_bea_year org_new  ) vce(cluster cluster1)  
+eststo clear
+
+g b1 = _b[x_p5]
+g b2 = _b[x_p4]
+g b3 = _b[x_p3]
+g b4 = _b[x_p2]
+g b5 = _b[x_p1]
+g b6 = _b[x]
+g b7 = _b[x_m1]
+g b8 = _b[x_m2]
+g b9 = _b[x_m3]
+g b10= _b[x_m4]
+g b11= _b[x_m5]
+g se1  =_se[x_p5]
+g se2  =_se[x_p4]
+g se3  =_se[x_p3]
+g se4  =_se[x_p2]
+g se5  =_se[x_p1]
+g se6  =_se[x]
+g se7  =_se[x_m1]
+g se8  =_se[x_m2]
+g se9  =_se[x_m3]
+g se10 =_se[x_m4]
+g se11 =_se[x_m5]
+
+
+lincom x_p5 
+g bb1  = r(estimate)
+g sse1 = r(se)
+lincom x_p5 + x_p4 
+g bb2  = r(estimate)
+g sse2 = r(se)
+lincom x_p5 + x_p4 + x_p3 
+g bb3  = r(estimate)
+g sse3 = r(se)
+lincom  x_p5 + x_p4 + x_p3 + x_p2 
+g bb4  = r(estimate)
+g sse4 = r(se)
+lincom x_p5 + x_p4 + x_p3 + x_p2 + x_p1 
+g bb5 = r(estimate)
+g sse5 = r(se)
+lincom x_p5 + x_p4 + x_p3 + x_p2 + x_p1 + x 
+g bb6  = r(estimate)
+g sse6 = r(se)
+lincom x_p5 + x_p4 + x_p3 + x_p2 + x_p1 + x + x_m1 
+g bb7  = r(estimate)
+g sse7 = r(se)
+lincom x_p5 + x_p4 + x_p3 + x_p2 + x_p1 + x + x_m1 + x_m2 
+g bb8  = r(estimate)
+g sse8 = r(se)
+lincom x_p5 + x_p4 + x_p3 + x_p2 + x_p1 + x + x_m1 + x_m2 + x_m3 
+g bb9  = r(estimate)
+g sse9 = r(se)
+lincom x_p5 + x_p4 + x_p3 + x_p2 + x_p1 + x + x_m1 + x_m2 + x_m3 + x_m4 
+g bb10  = r(estimate)
+g sse10 = r(se)
+lincom x_p5 + x_p4 + x_p3 + x_p2 + x_p1 + x + x_m1 + x_m2 + x_m3 + x_m4 + x_m5
+g bb11  = r(estimate)
+g sse11 = r(se)
+
+keep if _n==1
+keep b1-b11 se1-se11 bb* sse*
+save $main/data2/beta1_move, replace
+
+
+clear
+u $main/data2/beta1_move
+g n=_n
+reshape long b se bb sse, i(n) j(j)
+g t = j-6
+g upper = b + 1.96*se
+g lower = b - 1.96*se
+
+
+line  b upper lower t , legend(off) lpattern(solid dot  dot)  lcolor(black black black)     xlabel(-5 "{&beta}{sub:5}" -4 "{&beta}{sub:4}" -3 "{&beta}{sub:3}" -2 "{&beta}{sub:2}" -1 "{&beta}{sub:1}" 0 "{&beta}{sub:0}" 1 "{&beta}{sub:-1}" 2 "{&beta}{sub:-2}" 3 "{&beta}{sub:-3}" 4 "{&beta}{sub:-4}" 5 "{&beta}{sub:-5}" ) saving(fig1_m,replace)
+graph export fig1_m.pdf, replace
+! mv -f beta1_move.dta tables23/
+! mv -f fig1_m.pdf tables23/
+! mv -f fig1_m.gph tables23/
+
+drop upper lower
+g upper = bb + 1.96*sse
+g lower = bb - 1.96*sse
+
+
+
+line  bb upper lower t , legend(off) lpattern(solid dot  dot)  lcolor(black black black) yscale(range(-0.2 0.8)) ylabel(-0.2 0 0.2 0.4 0.6 0.8) xlabel(-5 "{&mu}{sub:5}" -4 "{&mu}{sub:4}" -3 "{&mu}{sub:3}" -2 "{&mu}{sub:2}" -1 "{&mu}{sub:1}" 0 "{&mu}{sub:0}" 1 "{&mu}{sub:-1}" 2 "{&mu}{sub:-2}" 3 "{&mu}{sub:-3}" 4 "{&mu}{sub:-4}" 5 "{&mu}{sub:-5}" ) saving(fig1b_m,replace)
+graph export fig1b_m.pdf, replace
+! mv -f fig1b_m.pdf tables23/
+! mv -f fig1b_m.gph tables23/
+
+
+
+restore
+
+

107/replication_package/AER_UPLOADED/reg27.do

@@ -0,0 +1,136 @@
+
+********************************************************************
+********************************************************************
+********************************************************************
+* KODAK - IO TABLES 
+********************************************************************
+********************************************************************
+********************************************************************
+
+u $main/data/read1
+summ
+summ value, detail
+rename class zd2
+sort zd
+save $main/data2/tmp27, replace
+
+
+
+
+********************************************************
+*****************************************
+*****************************************
+* PRODUCTIVITY 
+*****************************************
+*****************************************
+********************************************************
+clear
+u $main/data2/data_3 
+merge m:1 zd using $main/data2/tmp27
+tab _merge
+summ year zd value
+
+
+
+replace bea    =. if inter_bea   ==1
+replace zd     =. if inter_zd    ==1
+replace class  =. if inter_class ==1
+replace bea    =. if inter_bea2  ==1
+replace zd     =. if inter_zd2   ==1
+replace class  =. if inter_class2==1
+
+
+g y = log(number)
+g x   = log(Den_bea_zd    )
+
+egen cluster1               = group(bea zd)
+egen cluster_bea_year       = group(bea year)
+egen cluster_bea_zd_year    = group(bea zd year)
+egen cluster_class_year = group(bea class year)
+egen cluster_zd_year        = group(zd year)
+egen cluster_bea_class      = group(bea class)
+
+
+
+******************************************************************************
+* SAMPLE
+* I KEEP ONLY NON-KODAK INVENTORS
+* I KEEP ONLY 96 and 07
+******************************************************************************
+g kodak = 1 if org_id =="FI_1757"
+g xerox = 1 if org_id =="FI_6225"
+
+keep if y~=. & zd ~=. & class ~=. & year ~=. & inventor ~=. & cluster1 ~=. & bea ~=.
+keep if kodak ~=1
+drop if class ==396 | class ==399 
+keep if year ==1996 | year ==2007
+
+* DUMMIES
+****************
+* rochester dummy
+g roch = 0
+replace roch = 1 if bea ==139
+
+* dummy for 2007
+g       after = 0 if year ==1996
+replace after = 1 if year ==2007
+
+* Rochester * 2007
+g roch_after = roch*after
+
+
+
+*****************
+* CROSS-SECIONAL
+* DIFF IN DIFF
+*****************
+summ value, detail
+drop if value >0
+
+g cluster3 = bea
+preserve
+eststo:      reg y roch_after after roch,                                                      cluster(cluster3)  
+eststo: reghdfe  y roch_after after roch, absorb(zd)                                           cluster(cluster3) keepsin 
+eststo: reghdfe  y roch_after after roch, absorb(cluster_zd_year)                              cluster(cluster3) keepsin 
+eststo: reghdfe  y roch_after after roch, absorb(cluster1 cluster_zd_year)                     cluster(cluster3) keepsin 
+* drop Xerox inventors
+eststo: reghdfe  y roch_after after roch if xerox ~=1, absorb(cluster1 cluster_zd_year)                     cluster(cluster3) keepsin 
+
+
+
+
+**********************************************
+****************************************
+* WITHIN INVENTOR DIFFERENCES
+* I ONLY KEEP INVENTORS OBSERVED IN 1996 and 2007
+*****************************************
+**********************************************
+restore
+egen zzz = count(y), by(inventor)
+tab zzz
+keep if zzz ==2
+drop zzz
+preserve
+
+
+* New rochester dummy for rochester status in 1996
+* Note: if in Rochester in 1996, this dummy stays =1 both in 1996 and  2007
+drop roch roch_after
+g       tmp6 = 0
+replace tmp6 = 1 if bea ==139 & year == 1996
+egen roch = max(tmp6), by(inventor)
+g roch_after = roch*after
+
+eststo: reghdfe  y roch_after after roch, absorb(inventor)                           cluster(cluster3) keepsin
+eststo: reghdfe  y roch_after after roch, absorb(inventor zd)                        cluster(cluster3) keepsin
+eststo: reghdfe  y roch_after after roch, absorb(inventor cluster_zd_year)           cluster(cluster3) keepsin
+eststo: reghdfe  y roch_after after roch, absorb(inventor cluster_zd_year cluster1 ) cluster(cluster3) keepsin
+
+
+restore
+
+
+
+
+
+

107/replication_package/AER_UPLOADED/reg29.do

@@ -0,0 +1,215 @@
+
+********************************************************************
+********************************************************************
+********************************************************************
+* COUNTERFACTUAL WITH CURVATURE
+********************************************************************
+********************************************************************
+********************************************************************
+
+
+************************************
+***********************************
+* PRODUCTIVITY
+************************************
+***********************************
+clear
+u $main/data2/data_3
+egen total = sum(number), by(inventor)
+
+* INCLUDE THE FOLLOWING 5 LINES IF YOU DON'T WANT INTERPOLATED DATA
+replace bea    =. if inter_bea   ==1
+replace zd     =. if inter_zd    ==1
+replace class  =. if inter_class ==1
+replace bea    =. if inter_bea2  ==1
+replace zd     =. if inter_zd2   ==1
+replace class  =. if inter_class2==1
+
+********************************************************************
+********************************************************************
+********************************************************************
+********************************************************************
+********************************************************************
+********************************************************************
+
+************************
+************************
+* DEFINE  Y 
+***********************
+***********************
+* NUMBER OF PATENTS IN A YEAR
+g y = log(number)
+
+
+
+***********************
+***********************
+* DEFINE   X
+***********************
+***********************
+* NUMBER OF SCIENTISTS
+*g x    = Den_bea_zd
+*g x2   = Den_bea_class
+g x   = log(Den_bea_zd    )
+*g x2  = log(Den_bea_class )
+
+* NUMBER OF FIRMS
+*g x  = Den_bea_zdB 
+*g x2 = Den_bea_classB
+
+
+***********************
+***********************
+* DEFINE STAR INVENTORS
+***********************
+***********************
+keep if total  >=3.25
+
+
+
+*******************************************************************
+********************************************************************
+
+* THESE ARE THE 4 GROUPS USED IN THE ANALYSIS 
+* OF HETEROEGENITY BY CLUSTER SIZE
+summ x, detail
+egen pctile25 = pctile(x), p(25) by(zd2 year)
+egen pctile50 = pctile(x), p(50) by(zd2 year)
+egen pctile75 = pctile(x), p(75) by(zd2 year)
+g dd1 = 0
+g dd2 = 0
+g dd3 = 0
+g dd4 = 0
+replace dd1 =1 if x                <= pctile25
+replace dd2 =1 if x > pctile25 & x <= pctile50
+replace dd3 =1 if x > pctile50 & x <= pctile75
+replace dd4 =1 if x > pctile75
+
+
+
+
+
+*******************************************************************
+********************************************************************
+*******************************
+* COLLAPSE AND MERGE WITH DENSITY DATA THAT HAVE EVERY 
+* COMBINATION OF CITY-ZD-YEAR
+*
+* DATA HAVE ALL COMBINATION OF CITY-ZD-YEAR
+* I ADD 0's. THIS WAY I DO NOT DROP CELLS WITH NO PATENTS
+*
+*******************************
+*******************************
+*******************************************************************
+********************************************************************
+* COLLAPSE
+sort year bea_code zd2
+collapse y x numbe dd1 dd2 dd3 dd4, by(year bea_co zd2)
+summ
+
+tab dd1
+tab dd2 
+tab dd3
+tab dd4
+save tmp, replace
+
+* Merge with all combinations of bea_code zd2 year
+clear
+u tables2/data
+sort bea_code zd2 year
+merge 1:1  bea_code zd2 year using tmp
+drop _merge
+! rm tmp.dta
+rename density14 Den_bea_zd
+rename density14B Den_bea_zdB
+
+
+*egen minx = min(x)
+*egen miny = min(y)
+* I add the 0's
+replace number    = 0             if number    ==.
+*replace x         = 0.00001       if x         ==.
+*drop minx miny
+
+
+
+*******************************************************************
+********************************************************************
+*******************************
+* COUNTERFACTUALS 
+*******************************
+*******************************************************************
+********************************************************************
+drop _fillin Den_bea_zdB 
+
+
+
+* These are eintries  in the table on heterogenety by cluster size
+g       alpha = 0.0685 if dd1 ==1 
+replace alpha = 0.0743 if dd2 ==1
+replace alpha = 0.0720 if dd3 ==1
+replace alpha = 0.0776 if dd4 ==1
+
+
+
+
+* PERCENTILES
+egen tmp95 = pctile(Den_bea_zd), p(95) by(year zd2)
+egen tmp90 = pctile(Den_bea_zd), p(90) by(year zd2)
+egen tmp75 = pctile(Den_bea_zd), p(75) by(year zd2)
+egen tmp50 = pctile(Den_bea_zd), p(50) by(year zd2)
+egen tmp25 = pctile(Den_bea_zd), p(25) by(year zd2)
+
+
+* This is the counterfactual cluster size. 
+* I need to do the mean, so that the total 
+* number of scientsis in the US in teh 
+* counterfactual is = to observed number
+* I can't do median because it would change it
+egen mean = mean(Den_bea_zd), by(year zd2)
+
+
+* A log log model implies that the agglomeration effect in absolute amount of patents is cluster size to alpha
+* DD is estimated change in inventor productivity, measured in number of patents (not percent)
+g DD =   (mean^alpha) - (Den_bea_zd^alpha)
+* This is the percent change 
+g DDD = (mean^alpha - Den_bea_zd^alpha)*100/(Den_bea_zd^alpha)
+
+summ  DDD     if Den_bea_zd <=tmp25                       & Den_bea_zd ~=.
+summ  DDD     if Den_bea_zd >tmp25  & Den_bea_zd <= tmp50 & Den_bea_zd ~=.
+summ  DDD     if Den_bea_zd >tmp50  & Den_bea_zd <= tmp75 & Den_bea_zd ~=.
+summ  DDD     if Den_bea_zd >tmp75  & Den_bea_zd <= tmp90 & Den_bea_zd ~=.
+summ  DDD     if Den_bea_zd >tmp90  & Den_bea_zd <= tmp95 & Den_bea_zd ~=.
+summ  DDD     if Den_bea_zd >tmp95                        & Den_bea_zd ~=.
+drop tmp* 
+
+
+gsort year -Den_bea_zd
+*outfile  bea_name DDD     if year ==2007 & zd_name == "Semiconductors"        using tables2/table3b.txt, replace
+*outsheet bea_name DDD     if year ==2007 & zd_name == "Semiconductors"        using tables2/table3b.csv, replace comma
+*outfile  bea_name DDD     if year ==2007 & zd_name == "Biology and Chemestry" using tables2/table3c.txt, replace
+*outsheet bea_name DDD     if year ==2007 & zd_name == "Biology and Chemestry" using tables2/table3c.csv, replace comma
+*outfile  bea_name DDD     if year ==2007 & zd_name == "Computer Science"      using tables2/table3d.txt, replace
+*outsheet bea_name DDD     if year ==2007 & zd_name == "Computer Science"      using tables2/table3d.csv, replace comma
+
+
+* Change
+g ZZ =  Den_bea_zd*DD
+g ZZZ = Den_bea_zd*(Den_bea_zd^alpha)
+preserve
+collapse (sum)  ZZ ZZZ, by(year zd_name)
+
+* Aggregate effect by field
+* XXX TABLE 12 col 2
+g change = ZZ / ZZZ
+list zd change if year ==2007
+
+
+* Aggregate effect -- All fields
+* XXX TABLE 12 col 2
+restore
+collapse (sum) ZZ ZZZ, by(year)
+g change = ZZ / ZZZ
+list change if year ==2007
+
+

107/replication_package/AER_UPLOADED/reg3.do

@@ -0,0 +1,271 @@
+
+********************************************************************
+********************************************************************
+********************************************************************
+* KODAK 
+********************************************************************
+********************************************************************
+********************************************************************
+
+
+*******************************************************
+*****************************************
+*****************************************
+* GRAPHS: KODAK INVENTORS IN ROCHESTER
+*****************************************
+*****************************************
+*******************************************************
+* Number of inventors in Kodak in Rochester
+u $main/data2/data_patent_level_3
+g       kodak = 0
+replace kodak = 1 if org_id =="FI_1757"
+keep if kodak ==1 & bea_code ==139
+sort inventor_id year  
+by inventor_id year : g d1 = 1 if _n ==1
+egen count = count(d1), by(year)
+g kodak_tot = count
+
+collapse kodak_tot, by(year)
+
+keep if year >=1990
+line kodak_tot  year, ytitle("") xtitle("")
+graph export fig1.pdf, replace
+! mv -f fig1.pdf tables3/
+
+
+
+* Total number of inventors in Rochester
+clear
+u $main/data2/data_patent_level_3
+keep if bea_code ==139
+sort inventor_id year 
+by inventor_id year : g d1 = 1 if _n ==1
+egen count = count(d1), by(year)
+g rochester_tot = count
+collapse rochester_tot, by(year)
+keep if year >=1990
+list if year ==1996 | year ==2007
+
+
+
+********************************************************
+*********************************************
+****************************************
+* LIST OF PATENTING COMPANIES IN  ROCHESTER 
+*****************************************
+*********************************************
+********************************************************
+* Top patenters in Rochester outside Kodaak and Xerox in 1996
+clear
+u $main/data2/organizations_3
+keep if bea ==139 & year == 1996
+drop if org_id =="FI_1757" | org_id =="FI_6225"
+gsort  -org_N
+list org_norm org_N if org_N >10
+
+
+
+********************************************************
+*****************************************
+*****************************************
+* PRODUCTIVITY 
+*****************************************
+*****************************************
+********************************************************
+clear
+u $main/data2/data_3 
+
+replace bea    =. if inter_bea   ==1
+replace zd     =. if inter_zd    ==1
+replace class  =. if inter_class ==1
+replace bea    =. if inter_bea2  ==1
+replace zd     =. if inter_zd2   ==1
+replace class  =. if inter_class2==1
+
+
+g y = log(number)
+g x   = log(Den_bea_zd    )
+
+egen cluster1               = group(bea zd)
+egen cluster_bea_year       = group(bea year)
+egen cluster_bea_zd_year    = group(bea zd year)
+egen cluster_class_year = group(bea class year)
+egen cluster_zd_year        = group(zd year)
+egen cluster_bea_class      = group(bea class)
+
+
+
+******************************************************************************
+* SAMPLE
+* I KEEP ONLY NON-KODAK INVENTORS
+* I KEEP ONLY 96 and 07
+******************************************************************************
+g kodak = 1 if org_id =="FI_1757"
+g xerox = 1 if org_id =="FI_6225"
+
+keep if y~=. & zd ~=. & class ~=. & year ~=. & inventor ~=. & cluster1 ~=. & bea ~=.
+keep if kodak ~=1
+drop if class ==396 | class ==399 
+keep if year ==1996 | year ==2007
+
+* DUMMIES
+****************
+* rochester dummy
+g roch = 0
+replace roch = 1 if bea ==139
+
+* dummy for 2007
+g       after = 0 if year ==1996
+replace after = 1 if year ==2007
+
+* Rochester * 2007
+g roch_after = roch*after
+
+
+
+*****************
+* CROSS-SECIONAL
+* DIFF IN DIFF
+*****************
+g cluster3 = bea
+preserve
+eststo:      reg y roch_after after roch,                                                      cluster(cluster3)  
+eststo: reghdfe  y roch_after after roch, absorb(zd)                                           cluster(cluster3) keepsin 
+eststo: reghdfe  y roch_after after roch, absorb(cluster_zd_year)                              cluster(cluster3) keepsin 
+eststo: reghdfe  y roch_after after roch, absorb(cluster1 cluster_zd_year)                     cluster(cluster3) keepsin 
+
+esttab using tables3/table2.tex, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+esttab using tables3/table2, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+eststo clear
+
+
+
+***************************
+* 2SLS
+***************************
+* First stage Effect on cluster size
+eststo: reg x roch_after after roch,                                                      cluster(cluster3)
+test roch_after ==0
+eststo: reghdfe  x roch_after after roch, absorb(zd)                                           cluster(cluster3) keepsin
+test roch_after ==0
+eststo: reghdfe  x roch_after after roch, absorb(cluster_zd_year)                              cluster(cluster3) keepsin
+test roch_after ==0
+eststo: reghdfe  x roch_after after roch, absorb(cluster1 cluster_zd_year)                     cluster(cluster3) keepsin
+test roch_after ==0
+esttab using tables3/table5.tex, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+esttab using tables3/table5, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+eststo clear
+
+* IV
+eststo: ivreg     y (x =roch_after) after roch,                                  cluster(cluster3)
+eststo: ivreghdfe y (x =roch_after) after roch, absorb(zd                      ) cluster(cluster3)
+eststo: ivreghdfe y (x =roch_after) after roch, absorb(cluster_zd_year         ) cluster(cluster3)
+eststo: ivreghdfe y (x =roch_after) after roch, absorb(cluster1 cluster_zd_year) cluster(cluster3) 
+esttab using tables3/table6.tex, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+esttab using tables3/table6, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+eststo clear
+
+
+**********************************************
+****************************************
+* WITHIN INVENTOR DIFFERENCES
+* I ONLY KEEP INVENTORS OBSERVED IN 1996 and 2007
+*****************************************
+**********************************************
+restore
+egen zzz = count(y), by(inventor)
+tab zzz
+keep if zzz ==2
+drop zzz
+preserve
+
+
+* New rochester dummy for rochester status in 1996
+* Note: if in Rochester in 1996, this dummy stays =1 both in 1996 and  2007
+drop roch roch_after
+g       tmp6 = 0
+replace tmp6 = 1 if bea ==139 & year == 1996
+egen roch = max(tmp6), by(inventor)
+g roch_after = roch*after
+
+eststo: reghdfe  y roch_after after roch, absorb(inventor)                           cluster(cluster3) keepsin
+eststo: reghdfe  y roch_after after roch, absorb(inventor zd)                        cluster(cluster3) keepsin
+eststo: reghdfe  y roch_after after roch, absorb(inventor cluster_zd_year)           cluster(cluster3) keepsin
+eststo: reghdfe  y roch_after after roch, absorb(inventor cluster_zd_year cluster1 ) cluster(cluster3) keepsin
+
+esttab using tables3/table3.tex, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+esttab using tables3/table3, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+eststo clear
+
+
+* New rochester dummy for rochester status in 1996 and 2007
+* Note: it is =1 only of inventor is in rochester both in 1996 and  2007
+clear
+restore
+preserve
+drop roch roch_after
+
+g tmp6 = 1 if bea ==139 & year == 1996
+g tmp7 = 1 if bea ==139 & year == 2007
+egen tmp8 = max(tmp6), by(inventor)
+egen tmp9 = max(tmp7), by(inventor)
+g roch = 0
+replace roch = 1 if tmp8 ==1 & tmp9 ==1
+g roch_after = roch*after
+drop tmp*
+
+eststo: reghdfe  y roch_after after roch, absorb(inventor)                           cluster(cluster3) keepsin
+eststo: reghdfe  y roch_after after roch, absorb(inventor zd)                        cluster(cluster3) keepsin
+eststo: reghdfe  y roch_after after roch, absorb(inventor cluster_zd_year)           cluster(cluster3) keepsin
+eststo: reghdfe  y roch_after after roch, absorb(inventor cluster_zd_year cluster1 ) cluster(cluster3) keepsin
+
+esttab using tables3/table4.tex, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+esttab using tables3/table4, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+eststo clear
+
+
+
+
+******************************************************************************
+******************************************************************************
+* PRODUCTIVITY FIGURES
+******************************************************************************
+******************************************************************************
+clear
+u $main/data2/data_3
+
+replace bea    =. if inter_bea   ==1
+replace zd     =. if inter_zd    ==1
+replace class  =. if inter_class ==1
+replace bea    =. if inter_bea2  ==1
+replace zd     =. if inter_zd2   ==1
+replace class  =. if inter_class2==1
+
+g y = log(number)
+
+egen cluster1               = group(bea zd)
+g kodak = 1 if org_id =="FI_1757"
+keep if y~=. & zd ~=. & class ~=. & year ~=. & inventor ~=. & cluster1 ~=. & bea ~=.
+keep if kodak ~=1
+keep if bea_code ==139
+
+
+* RESIDUALIZED PRODUCTIVITY
+xi: areg y , absorb(class)
+predict fit3, xbd
+g e3 = y-fit3
+
+collapse y e3 , by(year bea)
+
+* FIGURES
+line e3  year if year >=1990 & bea ==139, xline(1996) yscale(r(-.3 .1))  ylabel(-.3 -.2 -.1 0 .1) ytitle("") xtitle("")
+graph export fig11.pdf, replace
+! mv -f fig11.pdf tables3/
+
+
+
+restore
+
+
+
+

107/replication_package/AER_UPLOADED/reg30.do

@@ -0,0 +1,161 @@
+********************************************************************
+********************************************************************
+********************************************************************
+* INTERPOLATION
+* log(1+patents)
+********************************************************************
+********************************************************************
+********************************************************************
+
+
+u $main/data2/data_3
+egen total = sum(number), by(inventor)
+
+
+* BY COMMENTING OUT THESE 6 LINES,
+* I DO NOT DROP INTERPOLATED DATA
+*replace bea    =. if inter_bea   ==1
+*replace zd     =. if inter_zd    ==1
+*replace class  =. if inter_class ==1
+*replace bea    =. if inter_bea2  ==1
+*replace zd     =. if inter_zd2   ==1
+*replace class  =. if inter_class2==1
+
+
+
+********************************************************************
+********************************************************************
+********************************************************************
+* OPTIONS
+********************************************************************
+********************************************************************
+********************************************************************
+
+************************
+************************
+* DEFINE  Y 
+***********************
+***********************
+* NUMBER OF PATENTS IN A YEAR
+g y = log(number)
+
+* CITATIONS
+*g y2 = log(citations+00001)
+g y2 = log(citations)
+
+* SPECIALIZATION
+*g y = 1-general
+
+
+***********************
+***********************
+* DEFINE   X
+***********************
+***********************
+* NUMBER OF SCIENTISTS
+*g x    = Den_bea_zd
+*g x2   = Den_bea_class
+g x   = log(Den_bea_zd    )
+g x2  = log(Den_bea_class )
+
+* NUMBER OF FIRMS
+*g x  = Den_bea_zdB 
+*g x2 = Den_bea_classB
+
+
+***********************
+***********************
+* DEFINE STAR INVENTORS
+***********************
+***********************
+*keep if total  >=3.25
+
+
+
+********************************************************************
+********************************************************************
+********************************************************************
+* NEW VARIABLES
+********************************************************************
+********************************************************************
+********************************************************************
+egen cluster1               = group(bea zd)
+egen cluster0               = group(bea zd year)
+egen cluster2               = group(bea )
+egen cluster_bea_year       = group(bea year)
+egen cluster_bea_class      = group(bea class)
+egen cluster_bea_zd_year    = group(bea zd year)
+egen cluster_bea_class_year = group(bea class year)
+egen cluster_zd_year        = group(zd year)
+egen cluster_class_year     = group(class year)
+egen org_new                = group(org_id)
+egen cluster_org_year       = group(org_new year)
+egen org_size               = count(inventor), by(org_new year)
+egen lab_size              = count(inventor), by(org_new year bea)
+
+sort inventor year
+by inventor : g  x_m1 = x[_n-1]
+by inventor : g  x_m2 = x[_n-2]
+by inventor : g  x_m3 = x[_n-3]
+by inventor : g  x_m4 = x[_n-4]
+by inventor : g  x_m5 = x[_n-5]
+by inventor : g  x_p1 = x[_n+1]
+by inventor : g  x_p2 = x[_n+2]
+by inventor : g  x_p3 = x[_n+3]
+by inventor : g  x_p4 = x[_n+4]
+by inventor : g  x_p5 = x[_n+5]
+by inventor : g  x2_m1 = x2[_n-1]
+by inventor : g  x2_m2 = x2[_n-2]
+by inventor : g  x2_m3 = x2[_n-3]
+by inventor : g  x2_m4 = x2[_n-4]
+by inventor : g  x2_m5 = x2[_n-5]
+by inventor : g  x2_p1 = x2[_n+1]
+by inventor : g  x2_p2 = x2[_n+2]
+by inventor : g  x2_p3 = x2[_n+3]
+by inventor : g  x2_p4 = x2[_n+4]
+by inventor : g  x2_p5 = x2[_n+5]
+
+
+
+
+
+
+********************************************************************
+********************************************************************
+********************************************************************
+* INTERPOLATED DATA
+********************************************************************
+********************************************************************
+********************************************************************
+
+* I ADD THE 0's IN INTERPOLATED YEARS
+drop y
+g y = log(number+1)
+replace y = log(1) if inter_bea ==1 | inter_zd ==1 | inter_class ==1
+replace y = log(1) if inter_bea2==1 | inter_zd2==1 | inter_class2==1
+
+
+sort inventor year
+by inventor: g  inter_org = 1   if org_new ==. & org_new[_n-1] == org_new[_n+1] & org_new[_n-1] ~=.
+by inventor: replace org_new = org_new[_n-1] if inter_org ==1
+
+by inventor: g  inter_org2 = 1  if org_new ==. & org_new[_n-1] ==. & org_new[_n-2] == org_new[_n+1] & org_new[_n-2] ~=.
+by inventor: replace org_new = org_new[_n-2] if inter_org2 ==1
+
+
+* This is the baseline. It shpould be equal to the main table with no interpolation
+eststo: reghdfe y x if total >=3.25 & inter_bea   ==0 & inter_zd ==0 & inter_class==0 & inter_bea2  ==0 & inter_zd2==0 & inter_class2==0 & inter_org ~=1 & inter_org2 ~=1, absorb(year bea zd class cluster_bea_year cluster_zd_year inventor cluster1 cluster_bea_class cluster_class_year org_new) cluster(cluster1) keepsin 
+
+* This regression uses interpolated data: 1 year
+eststo: reghdfe y x if total >=3.25 & inter_bea2  ==0 & inter_zd2==0 & inter_class2==0 & inter_org2 ~=1, absorb(year bea zd class cluster_bea_year cluster_zd_year inventor cluster1 cluster_bea_class cluster_class_year org_new) cluster(cluster1) keepsin 
+
+* This regression uses interpolated data: 2 year2
+eststo: reghdfe y x if total >=3.25, absorb(year bea zd class cluster_bea_year cluster_zd_year inventor cluster1 cluster_bea_class cluster_class_year org_new) cluster(cluster1) keepsin 
+
+esttab using tables30/table1.tex, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+esttab using tables30/table1.txt, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+eststo clear
+
+
+
+

107/replication_package/AER_UPLOADED/reg31.do

@@ -0,0 +1,167 @@
+********************************************************************
+********************************************************************
+********************************************************************
+* INTERPOLATION
+* inverse hyperbolic sine
+********************************************************************
+********************************************************************
+********************************************************************
+
+
+u $main/data2/data_3
+egen total = sum(number), by(inventor)
+
+
+* BY COMMENTING OUT THESE 6 LINES,
+* I DO NOT DROP INTERPOLATED DATA
+*replace bea    =. if inter_bea   ==1
+*replace zd     =. if inter_zd    ==1
+*replace class  =. if inter_class ==1
+*replace bea    =. if inter_bea2  ==1
+*replace zd     =. if inter_zd2   ==1
+*replace class  =. if inter_class2==1
+
+
+
+********************************************************************
+********************************************************************
+********************************************************************
+* OPTIONS
+********************************************************************
+********************************************************************
+********************************************************************
+
+************************
+************************
+* DEFINE  Y 
+***********************
+***********************
+* NUMBER OF PATENTS IN A YEAR
+g y = log(number)
+
+* CITATIONS
+*g y2 = log(citations+00001)
+g y2 = log(citations)
+
+* SPECIALIZATION
+*g y = 1-general
+
+
+***********************
+***********************
+* DEFINE   X
+***********************
+***********************
+* NUMBER OF SCIENTISTS
+*g x    = Den_bea_zd
+*g x2   = Den_bea_class
+g x   = log(Den_bea_zd    )
+g x2  = log(Den_bea_class )
+
+* NUMBER OF FIRMS
+*g x  = Den_bea_zdB 
+*g x2 = Den_bea_classB
+
+
+***********************
+***********************
+* DEFINE STAR INVENTORS
+***********************
+***********************
+*keep if total  >=3.25
+
+
+
+********************************************************************
+********************************************************************
+********************************************************************
+* NEW VARIABLES
+********************************************************************
+********************************************************************
+********************************************************************
+egen cluster1               = group(bea zd)
+egen cluster0               = group(bea zd year)
+egen cluster2               = group(bea )
+egen cluster_bea_year       = group(bea year)
+egen cluster_bea_class      = group(bea class)
+egen cluster_bea_zd_year    = group(bea zd year)
+egen cluster_bea_class_year = group(bea class year)
+egen cluster_zd_year        = group(zd year)
+egen cluster_class_year     = group(class year)
+egen org_new                = group(org_id)
+egen cluster_org_year       = group(org_new year)
+egen org_size               = count(inventor), by(org_new year)
+egen lab_size              = count(inventor), by(org_new year bea)
+
+sort inventor year
+by inventor : g  x_m1 = x[_n-1]
+by inventor : g  x_m2 = x[_n-2]
+by inventor : g  x_m3 = x[_n-3]
+by inventor : g  x_m4 = x[_n-4]
+by inventor : g  x_m5 = x[_n-5]
+by inventor : g  x_p1 = x[_n+1]
+by inventor : g  x_p2 = x[_n+2]
+by inventor : g  x_p3 = x[_n+3]
+by inventor : g  x_p4 = x[_n+4]
+by inventor : g  x_p5 = x[_n+5]
+by inventor : g  x2_m1 = x2[_n-1]
+by inventor : g  x2_m2 = x2[_n-2]
+by inventor : g  x2_m3 = x2[_n-3]
+by inventor : g  x2_m4 = x2[_n-4]
+by inventor : g  x2_m5 = x2[_n-5]
+by inventor : g  x2_p1 = x2[_n+1]
+by inventor : g  x2_p2 = x2[_n+2]
+by inventor : g  x2_p3 = x2[_n+3]
+by inventor : g  x2_p4 = x2[_n+4]
+by inventor : g  x2_p5 = x2[_n+5]
+
+
+
+
+
+
+********************************************************************
+********************************************************************
+********************************************************************
+* INTERPOLATED DATA
+********************************************************************
+********************************************************************
+********************************************************************
+
+* I ADD THE 0's IN INTERPOLATED YEARS
+drop y
+summ year number
+g y = asinh(number)
+summ year number y
+replace y = asinh(0) if inter_bea ==1 | inter_zd ==1 | inter_class ==1
+replace y = asinh(0) if inter_bea2==1 | inter_zd2==1 | inter_class2==1
+summ year number y
+summ year number y  if inter_bea ==1 | inter_zd ==1 | inter_class ==1
+summ year number y  if inter_bea2==1 | inter_zd2==1 | inter_class2==1
+
+
+sort inventor year
+by inventor: g  inter_org = 1   if org_new ==. & org_new[_n-1] == org_new[_n+1] & org_new[_n-1] ~=.
+by inventor: replace org_new = org_new[_n-1] if inter_org ==1
+
+by inventor: g  inter_org2 = 1  if org_new ==. & org_new[_n-1] ==. & org_new[_n-2] == org_new[_n+1] & org_new[_n-2] ~=.
+by inventor: replace org_new = org_new[_n-2] if inter_org2 ==1
+
+
+* This is the baseline. It shpould be equal to the main table with no interpolation
+eststo: reghdfe y x if total >=3.25 & inter_bea   ==0 & inter_zd ==0 & inter_class==0 & inter_bea2  ==0 & inter_zd2==0 & inter_class2==0 & inter_org ~=1 & inter_org2 ~=1, absorb(year bea zd class cluster_bea_year cluster_zd_year inventor cluster1 cluster_bea_class cluster_class_year org_new) cluster(cluster1) keepsin 
+
+* This regression uses interpolated data: 1 year
+eststo: reghdfe y x if total >=3.25 & inter_bea2  ==0 & inter_zd2==0 & inter_class2==0 & inter_org2 ~=1, absorb(year bea zd class cluster_bea_year cluster_zd_year inventor cluster1 cluster_bea_class cluster_class_year org_new) cluster(cluster1) keepsin 
+
+* This regression uses interpolated data: 2 year2
+eststo: reghdfe y x if total >=3.25, absorb(year bea zd class cluster_bea_year cluster_zd_year inventor cluster1 cluster_bea_class cluster_class_year org_new) cluster(cluster1) keepsin 
+
+esttab using tables31/table1.tex, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+esttab using tables31/table1.txt, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+eststo clear
+
+
+
+
+

107/replication_package/AER_UPLOADED/reg34.do

@@ -0,0 +1,194 @@
+********************************************************************
+********************************************************************
+********************************************************************
+* DIFFERENT TIME UNIT (1,2,3,6 MONTHS, 2, 3 YEAR)  -- SUPERSTARS
+********************************************************************
+********************************************************************
+********************************************************************
+
+
+
+*****************
+* 1 MONTH
+*****************
+clear
+u $main/data2/data_4a
+egen total = sum(number), by(inventor)
+
+g y = log(number)
+g x   = log(Den_bea_zd    )
+
+egen cluster1               = group(bea zd)
+egen cluster_bea_year       = group(bea year)
+egen cluster_bea_zd_year    = group(bea zd year)
+egen cluster_zd_year        = group(zd year)
+egen cluster_class_year     = group(class year)
+egen cluster_bea_class      = group(bea class)
+egen org_new2               = group(org_new)
+drop org_new
+
+keep if total  >=13.6
+keep if y~=. & zd ~=. & class ~=. & year ~=. & inventor ~=. & cluster1 ~=. & bea ~=.
+summ y x 
+
+eststo: reghdfe y x , absorb(year bea zd class cluster_bea_year cluster_zd_year cluster_class_year inventor cluster1 cluster_bea_class cluster_class_year org_new) vce(cluster cluster1) keepsin 
+
+
+*****************
+* 2 MONTHS
+*****************
+clear
+u $main/data2/data_4e
+egen total = sum(number), by(inventor)
+
+g y = log(number)
+g x   = log(Den_bea_zd    )
+
+egen cluster1               = group(bea zd)
+egen cluster_bea_year       = group(bea year)
+egen cluster_bea_zd_year    = group(bea zd year)
+egen cluster_zd_year        = group(zd year)
+egen cluster_class_year     = group(class year)
+egen cluster_bea_class      = group(bea class)
+egen org_new2               = group(org_new)
+drop org_new
+
+keep if total  >=13.6
+keep if y~=. & zd ~=. & class ~=. & year ~=. & inventor ~=. & cluster1 ~=. & bea ~=.
+summ y x
+
+eststo: reghdfe y x , absorb(year bea zd class cluster_bea_year cluster_zd_year cluster_class_year inventor cluster1 cluster_bea_class cluster_class_year org_new) vce(cluster cluster1)  keepsin 
+
+*****************
+* 3 MONTH
+*****************
+clear
+u $main/data2/data_4f
+egen total = sum(number), by(inventor)
+
+g y = log(number)
+g x   = log(Den_bea_zd    )
+
+egen cluster1               = group(bea zd)
+egen cluster_bea_year       = group(bea year)
+egen cluster_bea_zd_year    = group(bea zd year)
+egen cluster_zd_year        = group(zd year)
+egen cluster_class_year     = group(class year)
+egen cluster_bea_class      = group(bea class)
+egen org_new2               = group(org_new)
+drop org_new
+
+keep if total  >=13.6
+keep if y~=. & zd ~=. & class ~=. & year ~=. & inventor ~=. & cluster1 ~=. & bea ~=.
+summ y x
+
+eststo: reghdfe y x , absorb(year bea zd class cluster_bea_year cluster_zd_year cluster_class_year inventor cluster1 cluster_bea_class cluster_class_year org_new) vce(cluster cluster1)  keepsin 
+
+
+*****************
+* 6 MONTH
+*****************
+clear
+u $main/data2/data_4b
+egen total = sum(number), by(inventor)
+
+g y = log(number)
+g x   = log(Den_bea_zd    )
+
+egen cluster1               = group(bea zd)
+egen cluster_bea_year       = group(bea year)
+egen cluster_bea_zd_year    = group(bea zd year)
+egen cluster_zd_year        = group(zd year)
+egen cluster_class_year     = group(class year)
+egen cluster_bea_class      = group(bea class)
+egen org_new2               = group(org_new)
+drop org_new
+
+
+keep if total  >=13.6
+keep if y~=. & zd ~=. & class ~=. & year ~=. & inventor ~=. & cluster1 ~=. & bea ~=.
+summ y x 
+
+eststo: reghdfe y x , absorb(year bea zd class cluster_bea_year cluster_zd_year cluster_class_year inventor cluster1 cluster_bea_class cluster_class_year org_new) vce(cluster cluster1) keepsin 
+
+*****************
+* 1 YEAR
+*****************
+clear
+u $main/data2/data_3
+egen total = sum(number), by(inventor)
+
+g y = log(number)
+g x   = log(Den_bea_zd    )
+
+egen cluster1               = group(bea zd)
+egen cluster_bea_year       = group(bea year)
+egen cluster_bea_zd_year    = group(bea zd year)
+egen cluster_zd_year        = group(zd year)
+egen cluster_class_year     = group(class year)
+egen cluster_bea_class      = group(bea class)
+egen org_new                = group(org_id)
+
+keep if total  >=13.6
+keep if y~=. & zd ~=. & class ~=. & year ~=. & inventor ~=. & cluster1 ~=. & bea ~=.
+summ y x 
+
+eststo: reghdfe y x , absorb(year bea zd class cluster_bea_year cluster_zd_year cluster_class_year inventor cluster1 cluster_bea_class cluster_class_year org_new) vce(cluster cluster1) keepsin 
+
+*****************
+* 2 YEAR
+*****************
+clear
+u $main/data2/data_4c
+egen total = sum(number), by(inventor)
+
+g y = log(number)
+g x   = log(Den_bea_zd    )
+
+egen cluster1               = group(bea zd)
+egen cluster_bea_year       = group(bea year)
+egen cluster_bea_zd_year    = group(bea zd year)
+egen cluster_zd_year        = group(zd year)
+egen cluster_class_year     = group(class year)
+egen cluster_bea_class      = group(bea class)
+egen org_new2               = group(org_new)
+drop org_new
+
+keep if total  >=13.6
+keep if y~=. & zd ~=. & class ~=. & year ~=. & inventor ~=. & cluster1 ~=. & bea ~=.
+summ y x 
+
+eststo: reghdfe y x , absorb(year bea zd class cluster_bea_year cluster_zd_year cluster_class_year inventor cluster1 cluster_bea_class cluster_class_year org_new) vce(cluster cluster1) keepsin 
+
+*****************
+* 3 YEAR
+*****************
+clear
+u $main/data2/data_4d
+egen total = sum(number), by(inventor)
+
+g y = log(number)
+g x   = log(Den_bea_zd    )
+
+egen cluster1               = group(bea zd)
+egen cluster_bea_year       = group(bea year)
+egen cluster_bea_zd_year    = group(bea zd year)
+egen cluster_zd_year        = group(zd year)
+egen cluster_class_year     = group(class year)
+egen cluster_bea_class      = group(bea class)
+egen org_new2               = group(org_new)
+drop org_new
+
+
+keep if total  >=13.6
+keep if y~=. & zd ~=. & class ~=. & year ~=. & inventor ~=. & cluster1 ~=. & bea ~=.
+summ y x 
+
+eststo: reghdfe y x , absorb(year bea zd class cluster_bea_year cluster_zd_year cluster_class_year inventor cluster1 cluster_bea_class cluster_class_year org_new) vce(cluster cluster1) keepsin 
+esttab using tables34/table9.tex, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+esttab using tables34/table9.txt, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+eststo clear
+
+
+*restore
+

107/replication_package/AER_UPLOADED/reg36.do

@@ -0,0 +1,277 @@
+********************************************************************
+********************************************************************
+********************************************************************
+* IV in appendix
+********************************************************************
+********************************************************************
+********************************************************************
+
+
+********************************************************************
+********************************************************************
+********************************************************************
+* READ DATA
+********************************************************************
+********************************************************************
+********************************************************************
+clear
+u $main/data2/data_3
+
+egen total = sum(number), by(inventor)
+
+* INCLUDE THE FOLLOWING 6 LINES IF YOU DON'T WANT INTERPOLATED DATA
+replace bea    =. if inter_bea   ==1
+replace zd     =. if inter_zd    ==1
+replace class  =. if inter_class ==1
+replace bea    =. if inter_bea2  ==1
+replace zd     =. if inter_zd2   ==1
+replace class  =. if inter_class2==1
+
+
+
+* CLUSTER SIZE 
+g x   = log(Den_bea_zd    )
+sort inventor year
+by inventor : g  x_m1 = x[_n-1]
+by inventor : g  x_m2 = x[_n-2]
+by inventor : g  x_m3 = x[_n-3]
+by inventor : g  x_m4 = x[_n-4]
+by inventor : g  x_m5 = x[_n-5]
+by inventor : g  x_p1 = x[_n+1]
+by inventor : g  x_p2 = x[_n+2]
+by inventor : g  x_p3 = x[_n+3]
+by inventor : g  x_p4 = x[_n+4]
+by inventor : g  x_p5 = x[_n+5]
+
+
+
+
+drop if year  ==.
+drop if zd ==.
+drop if bea ==.
+egen org_new2 = group(org_id) if org_id ~= ""
+
+
+
+********************************************************************
+********************************************************************
+********************************************************************
+********************************************************************
+********************************************************************
+********************************************************************
+
+* NUMBER OF PATENTS IN A YEAR
+g y = log(number)
+
+* CITATIONS
+* To avoid dropping the 0's, I add 0.00001
+g y2 = log(citations+0.00001)
+g y3 = log( (citations+0.00001) / number) 
+
+* SPECIALIZATION
+*g y4 = 1-general
+
+* NUMBER OF FIRMS
+g x3  = log(Den_bea_zdB) 
+
+
+***********************
+***********************
+* DEFINE STAR INVENTORS
+* top 10% = 3.25
+***********************
+***********************
+keep if total  >=3.25
+
+
+
+********************************************************************
+********************************************************************
+********************************************************************
+* NEW VARIABLES
+********************************************************************
+********************************************************************
+********************************************************************
+egen cluster1               = group(bea zd)
+egen cluster0               = group(bea zd year)
+egen cluster2               = group(bea )
+egen cluster_bea_year       = group(bea year)
+egen cluster_bea_zd_year    = group(bea zd year)
+egen cluster_bea_class_year = group(bea class year)
+egen cluster_zd_year        = group(zd year)
+egen cluster_bea_class      = group(bea class)
+egen cluster_class_year     = group(class year)
+egen org_new                = group(org_id)
+egen cluster_org_year       = group(org_new year)
+egen org_size               = count(inventor), by(org_new year)
+egen lab_size              = count(inventor), by(org_new year bea)
+
+
+
+********************************************************************
+********************************************************************
+********************************************************************
+* REGRESSIONS
+********************************************************************
+********************************************************************
+********************************************************************
+keep if y~=. & zd ~=. & class ~=. & year ~=. & inventor ~=. & cluster1 ~=. & bea ~=.
+summ y y2 y3  x 
+summ number, detail
+summ x Den_bea_zd , detail
+tab zd
+
+
+
+
+
+***************************************
+*MODELS IN DIFFERENCES and IV
+***************************************
+save tmp35, replace
+
+
+* Now I indetify firms that have a presence in only 1 city
+sort bea year org_new2
+g jjj=1 if org_new2 ~= .
+collapse jjj, by(bea year org_new2)
+
+* number of cities a firm is present in a given year
+egen NN2 = sum(jjj), by(year org_new2)
+* max number of cities a firm has ever been present 
+egen NN3 = max(NN2), by(org_new2)
+summ NN2 NN3, detail
+sort bea year org_new2
+save tmp89, replace
+
+
+************************************
+
+clear
+u tmp35
+
+drop if org_id == ""
+sort org_new2 bea year zd
+merge m:1   bea year zd org_new2 using  $main/data2/iv_data_new
+drop _merge
+
+sort bea year org_new2
+merge m:1  bea year org_new2 using tmp89
+drop _merge
+rm tmp89.dta
+rm tmp35.dta
+
+g IV = iv8
+
+sort inventor year
+by inventor: g Dy    = y - y[_n-1]
+by inventor: g Dx    = x - x[_n-1]
+by inventor: g Dyear = year - year[_n-1]
+by inventor: g Dbea  = 1 if bea_code == bea_code[_n-1] 
+by inventor: g Dy_2    = y - y[_n-2]
+by inventor: g Dx_2    = x - x[_n-2]
+by inventor: g Dyear_2 = year - year[_n-2]
+by inventor: g Dbea_2  = 1 if bea_code == bea_code[_n-2]
+by inventor: g Dx_m1 = x[_n-1] - x[_n-2]
+by inventor: g Dx_m2 = x[_n-2] - x[_n-3]
+by inventor: g Dx_m3 = x[_n-3] - x[_n-2]
+egen cluster10       = group(bea )
+
+
+* Models in first differences should only include observations in contiguous years
+keep if Dyear ==1
+keep if Dy ~=. & Dx ~=. & IV ~=.
+
+
+drop if NN3 >1
+
+
+* OLS
+eststo: reghdfe  Dy Dx, absorb(year  zd class )  vce(cluster cluster10)
+eststo: reghdfe  Dy Dx, absorb(year  zd class cluster_zd_year                    )  vce(cluster cluster10)
+eststo: reghdfe  Dy Dx, absorb(year  zd class cluster_zd_year cluster_class_year )  vce(cluster cluster10)
+* 2SLS
+eststo: ivreghdfe Dy (Dx =IV), absorb(year zd class ) cluster(cluster10)
+eststo: ivreghdfe Dy (Dx =IV), absorb(year zd class cluster_zd_year ) cluster(cluster10)
+eststo: ivreghdfe Dy (Dx =IV), absorb(year zd class cluster_zd_year cluster_class_year ) cluster(cluster10)
+
+esttab using tables36/table22.tex, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+esttab using tables36/table22, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+eststo clear
+
+* First Stage
+eststo: reghdfe Dx IV, absorb(year zd class ) vce(cluster cluster10) res(v4)
+test IV=0
+eststo: reghdfe Dx IV, absorb(year zd class cluster_zd_year ) vce(cluster cluster10) res(v5) 
+test IV=0
+eststo: reghdfe Dx IV, absorb(year zd class cluster_zd_year cluster_class_year ) vce(cluster cluster10) res(v6)
+test IV=0
+
+esttab using tables36/table21.tex, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+esttab using tables36/table21, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+eststo clear
+
+
+* EXOGENEITY TESTS
+reghdfe  Dy Dx v4, absorb(year  zd class )  vce(cluster cluster10)
+test v4
+reghdfe  Dy Dx v5, absorb(year  zd class cluster_zd_year                    )  vce(cluster cluster10)
+test v5
+reghdfe  Dy Dx v6, absorb(year  zd class cluster_zd_year cluster_class_year )  vce(cluster cluster10)
+test v6
+drop v4 v5 v6
+
+
+
+***************************************
+* SAME CITY AND FIELD
+egen max = max(bea), by(inventor)
+egen min = min(bea), by(inventor)
+keep if max == min
+drop max min
+
+egen max = max(zd), by(inventor)
+egen min = min(zd), by(inventor)
+keep if max == min
+drop max min
+
+
+* OLS
+eststo: reghdfe  Dy Dx, absorb(year  zd class )  vce(cluster cluster10)
+eststo: reghdfe  Dy Dx, absorb(year  zd class cluster_zd_year                    )  vce(cluster cluster10)
+eststo: reghdfe  Dy Dx, absorb(year  zd class cluster_zd_year cluster_class_year )  vce(cluster cluster10)
+* 2SLS
+eststo: ivreghdfe Dy (Dx =IV), absorb(year zd class ) cluster(cluster10)
+eststo: ivreghdfe Dy (Dx =IV), absorb(year zd class cluster_zd_year ) cluster(cluster10)
+eststo: ivreghdfe Dy (Dx =IV), absorb(year zd class cluster_zd_year cluster_class_year ) cluster(cluster10)
+
+esttab using tables36/table222.tex, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+esttab using tables36/table222, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+eststo clear
+
+* First Stage
+eststo: reghdfe Dx IV, absorb(year zd class ) vce(cluster cluster10) res(v4)
+test IV=0
+eststo: reghdfe Dx IV, absorb(year zd class cluster_zd_year ) vce(cluster cluster10) res(v5)
+test IV=0
+eststo: reghdfe Dx IV, absorb(year zd class cluster_zd_year cluster_class_year ) vce(cluster cluster10) res(v6)
+test IV=0
+
+esttab using tables36/table221.tex, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+esttab using tables36/table221, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+eststo clear
+
+
+
+* EXOGENEITY TESTS
+reghdfe  Dy Dx v4, absorb(year  zd class )  vce(cluster cluster10)
+test v4
+reghdfe  Dy Dx v5, absorb(year  zd class cluster_zd_year                    )  vce(cluster cluster10)
+test v5
+reghdfe  Dy Dx v6, absorb(year  zd class cluster_zd_year cluster_class_year )  vce(cluster cluster10)
+test v6
+
+
+
+restore
+

107/replication_package/AER_UPLOADED/reg4.do

@@ -0,0 +1,480 @@
+********************************************************************
+********************************************************************
+********************************************************************
+* ROBUSTNESS:
+* INTERPOLATION; DIFFERENT AMOUNTS OF CENSORING; LAGGED X; 
+* DIFFERENT TIME UNIT (1,2,3,6 MONTHS, 2, 3 YEAR) 
+********************************************************************
+********************************************************************
+********************************************************************
+
+
+u $main/data2/data_3
+egen total = sum(number), by(inventor)
+
+
+* BY COMMENTING OUT THESE 6 LINES,
+* I DO NOT DROP INTERPOLATED DATA
+*replace bea    =. if inter_bea   ==1
+*replace zd     =. if inter_zd    ==1
+*replace class  =. if inter_class ==1
+*replace bea    =. if inter_bea2  ==1
+*replace zd     =. if inter_zd2   ==1
+*replace class  =. if inter_class2==1
+
+
+
+********************************************************************
+********************************************************************
+********************************************************************
+* OPTIONS
+********************************************************************
+********************************************************************
+********************************************************************
+
+************************
+************************
+* DEFINE  Y 
+***********************
+***********************
+* NUMBER OF PATENTS IN A YEAR
+g y = log(number)
+
+* CITATIONS
+*g y2 = log(citations+00001)
+g y2 = log(citations)
+
+* SPECIALIZATION
+*g y = 1-general
+
+
+***********************
+***********************
+* DEFINE   X
+***********************
+***********************
+* NUMBER OF SCIENTISTS
+*g x    = Den_bea_zd
+*g x2   = Den_bea_class
+g x   = log(Den_bea_zd    )
+g x2  = log(Den_bea_class )
+
+* NUMBER OF FIRMS
+*g x  = Den_bea_zdB 
+*g x2 = Den_bea_classB
+
+
+***********************
+***********************
+* DEFINE STAR INVENTORS
+***********************
+***********************
+*keep if total  >=3.25
+
+
+
+********************************************************************
+********************************************************************
+********************************************************************
+* NEW VARIABLES
+********************************************************************
+********************************************************************
+********************************************************************
+egen cluster1               = group(bea zd)
+egen cluster0               = group(bea zd year)
+egen cluster2               = group(bea )
+egen cluster_bea_year       = group(bea year)
+egen cluster_bea_class      = group(bea class)
+egen cluster_bea_zd_year    = group(bea zd year)
+egen cluster_bea_class_year = group(bea class year)
+egen cluster_zd_year        = group(zd year)
+egen cluster_class_year     = group(class year)
+egen org_new                = group(org_id)
+egen cluster_org_year       = group(org_new year)
+egen org_size               = count(inventor), by(org_new year)
+egen lab_size              = count(inventor), by(org_new year bea)
+
+sort inventor year
+by inventor : g  x_m1 = x[_n-1]
+by inventor : g  x_m2 = x[_n-2]
+by inventor : g  x_m3 = x[_n-3]
+by inventor : g  x_m4 = x[_n-4]
+by inventor : g  x_m5 = x[_n-5]
+by inventor : g  x_p1 = x[_n+1]
+by inventor : g  x_p2 = x[_n+2]
+by inventor : g  x_p3 = x[_n+3]
+by inventor : g  x_p4 = x[_n+4]
+by inventor : g  x_p5 = x[_n+5]
+by inventor : g  x2_m1 = x2[_n-1]
+by inventor : g  x2_m2 = x2[_n-2]
+by inventor : g  x2_m3 = x2[_n-3]
+by inventor : g  x2_m4 = x2[_n-4]
+by inventor : g  x2_m5 = x2[_n-5]
+by inventor : g  x2_p1 = x2[_n+1]
+by inventor : g  x2_p2 = x2[_n+2]
+by inventor : g  x2_p3 = x2[_n+3]
+by inventor : g  x2_p4 = x2[_n+4]
+by inventor : g  x2_p5 = x2[_n+5]
+
+
+
+
+
+
+********************************************************************
+********************************************************************
+********************************************************************
+* INTERPOLATED DATA
+* NOTE:  This analysis is out of date. 
+* log(0) is not defined. 0's are transformed using log(0.001)
+* More recent versions of this table are in reg30.do and reg31.do
+********************************************************************
+********************************************************************
+********************************************************************
+
+* I ADD THE 0's IN INTERPOLATED YEARS
+replace y = log(0.001) if inter_bea ==1 | inter_zd ==1 | inter_class ==1
+replace y = log(0.001) if inter_bea2==1 | inter_zd2==1 | inter_class2==1
+
+
+sort inventor year
+by inventor: g  inter_org = 1   if org_new ==. & org_new[_n-1] == org_new[_n+1] & org_new[_n-1] ~=.
+by inventor: replace org_new = org_new[_n-1] if inter_org ==1
+
+by inventor: g  inter_org2 = 1  if org_new ==. & org_new[_n-1] ==. & org_new[_n-2] == org_new[_n+1] & org_new[_n-2] ~=.
+by inventor: replace org_new = org_new[_n-2] if inter_org2 ==1
+
+
+* This is the baseline. It shpould be equal to the main table with no interpolation
+eststo: reghdfe y x if total >=3.25 & inter_bea   ==0 & inter_zd ==0 & inter_class==0 & inter_bea2  ==0 & inter_zd2==0 & inter_class2==0 & inter_org ~=1 & inter_org2 ~=1, absorb(year bea zd class cluster_bea_year cluster_zd_year inventor cluster1 cluster_bea_class cluster_class_year org_new) cluster(cluster1) keepsin 
+
+* This regression uses interpolated data: 1 year
+eststo: reghdfe y x if total >=3.25 & inter_bea2  ==0 & inter_zd2==0 & inter_class2==0 & inter_org2 ~=1, absorb(year bea zd class cluster_bea_year cluster_zd_year inventor cluster1 cluster_bea_class cluster_class_year org_new) cluster(cluster1) keepsin 
+
+* This regression uses interpolated data: 2 year2
+eststo: reghdfe y x if total >=3.25, absorb(year bea zd class cluster_bea_year cluster_zd_year inventor cluster1 cluster_bea_class cluster_class_year org_new) cluster(cluster1) keepsin 
+
+esttab using tables4/table1.tex, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+esttab using tables4/table1.txt, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+eststo clear
+
+
+********************************************************************
+********************************************************************
+********************************************************************
+* DIFFERENT DEFINITIONS OF STARS
+* (USING NON INTERPOLATED DATA)
+********************************************************************
+********************************************************************
+********************************************************************
+
+* DROP INTERPOLATED OBS
+drop if inter_bea  ==1
+drop if inter_zd    ==1
+drop if inter_class ==1
+drop if inter_bea2 ==1
+drop if inter_zd2   ==1
+drop if inter_class2==1
+
+summ y x x2
+
+eststo: reghdfe y x              ,    absorb(year bea zd class cluster_bea_year cluster_zd_year inventor cluster1 cluster_bea_class cluster_class_year org_new) cluster(cluster1)     keepsin 
+eststo: reghdfe y x if total  >=1.5,  absorb(year bea zd class cluster_bea_year cluster_zd_year inventor cluster1 cluster_bea_class cluster_class_year org_new) cluster(cluster1)   keepsin 
+eststo: reghdfe y x if total  >=3.25, absorb(year bea zd class cluster_bea_year cluster_zd_year inventor cluster1 cluster_bea_class cluster_class_year org_new) cluster(cluster1)  keepsin 
+eststo: reghdfe y x if total  >=5.4,  absorb(year bea zd class cluster_bea_year cluster_zd_year inventor cluster1 cluster_bea_class cluster_class_year org_new) cluster(cluster1)   keepsin 
+eststo: reghdfe y x if total  >=13.6, absorb(year bea zd class cluster_bea_year cluster_zd_year inventor cluster1 cluster_bea_class cluster_class_year org_new) cluster(cluster1)  keepsin 
+eststo: reghdfe y x if total  >=15,   absorb(year bea zd class cluster_bea_year cluster_zd_year inventor cluster1 cluster_bea_class cluster_class_year org_new) cluster(cluster1)    keepsin 
+esttab using tables4/table2.tex, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+esttab using tables4/table2.txt, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+
+eststo clear
+
+* Average number of years  
+egen n_year = count(y), by(inventor) 
+summ total n_year
+drop n_year
+egen n_year = count(y) if total  >=1.5, by(inventor)
+summ total n_year
+drop n_year
+egen n_year = count(y) if total  >=3.25, by(inventor)
+summ total n_year
+drop n_year
+egen n_year = count(y) if total  >=5.4, by(inventor)
+summ total n_year
+drop n_year
+egen n_year = count(y) if total  >=13.6, by(inventor)
+summ total n_year
+drop n_year
+egen n_year = count(y) if total  >=15, by(inventor)
+summ total n_year
+drop n_year
+
+
+
+
+
+********************************************************************
+********************************************************************
+********************************************************************
+* TO AVOID MECHANICAL CORRELATION
+* ASSIGN CLUSTER SIZE THAT EXISTED IN THE RELEVANT BEA-ZD LAST YEAR 
+* OR 2 YEARS AGO. 
+* NOTE: FOR A GIVEN INVENTOR, IT IS NOT 
+*       THE LAGGED CLUSTER SIZE DUE TO MOBILITY
+********************************************************************
+********************************************************************
+********************************************************************
+preserve
+
+* I MERGE CLUSTER DENSITY 1 YEAR AGO
+g year2 = year
+replace year = year -1
+sort bea zd year
+merge m:1 bea zd year using data2/density14_3
+tab _merge
+drop _merge
+g Den_bea_zd_1 = density14
+drop density14*
+
+* I MERGE CLUSTER DENSITY 2 YEARS AGO
+drop year
+g year = year2-2
+sort bea zd year
+merge m:1 bea zd year using data2/density14_3
+tab _merge
+drop _merge
+g Den_bea_zd_2 = density14
+drop density14*
+drop year
+
+* SERIAL CORRELATION IN X
+rename year2 year
+g xx   = log(Den_bea_zd_1)
+g xxx  = log(Den_bea_zd_2)
+
+correlate(x xx)
+correlate(x xxx)
+reg x xx
+reg x xxx
+
+egen xxxx   = rowmean(x xx xxx)
+egen xxxxx  = rowmin(x xx xxx)
+egen xxxxxx = rowmax(x xx xxx)
+
+
+keep if total >=3.25
+
+* Baseline
+eststo: reghdfe y x ,  absorb(year bea zd class cluster_bea_year cluster_zd_year inventor cluster1 cluster_bea_class cluster_class_year org_new) cluster(cluster1)  keepsin 
+
+* Using x that existed in the same cluster in the previous year 
+eststo: reghdfe y xx,  absorb(year bea zd class cluster_bea_year cluster_zd_year inventor cluster1 cluster_bea_class cluster_class_year org_new) cluster(cluster1)  keepsin 
+
+* Using x that existed in the same cluster 2 years before
+eststo: reghdfe y xxx, absorb(year bea zd class cluster_bea_year cluster_zd_year inventor cluster1 cluster_bea_class cluster_class_year org_new) cluster(cluster1)  keepsin 
+
+* Using mean, min or max over the last 3 years
+eststo: reghdfe y xxxx,   absorb(year bea zd class cluster_bea_year cluster_zd_year inventor cluster1 cluster_bea_class cluster_class_year org_new) cluster(cluster1) keepsin 
+eststo: reghdfe y xxxxx,  absorb(year bea zd class cluster_bea_year cluster_zd_year inventor cluster1 cluster_bea_class cluster_class_year org_new) cluster(cluster1) keepsin 
+eststo: reghdfe y xxxxxx, absorb(year bea zd class cluster_bea_year cluster_zd_year inventor cluster1 cluster_bea_class cluster_class_year org_new) cluster(cluster1) keepsin 
+
+esttab using tables4/table3.tex, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+esttab using tables4/table3.txt, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+
+eststo clear
+
+
+
+
+********************************************************************
+********************************************************************
+********************************************************************
+* 1 MONTH; 2 MONTHS; 3 MONTHS; 6 MONTHS; 2 YEARS; 3 YEARS
+********************************************************************
+********************************************************************
+********************************************************************
+
+*****************
+* 1 MONTH
+*****************
+clear
+u $main/data2/data_4a
+egen total = sum(number), by(inventor)
+
+g y = log(number)
+g x   = log(Den_bea_zd    )
+
+egen cluster1               = group(bea zd)
+egen cluster_bea_year       = group(bea year)
+egen cluster_bea_zd_year    = group(bea zd year)
+egen cluster_zd_year        = group(zd year)
+egen cluster_class_year     = group(class year)
+egen cluster_bea_class      = group(bea class)
+egen org_new2               = group(org_new)
+drop org_new
+
+keep if total  >=3.25
+keep if y~=. & zd ~=. & class ~=. & year ~=. & inventor ~=. & cluster1 ~=. & bea ~=.
+summ y x 
+
+eststo: reghdfe y x , absorb(year bea zd class cluster_bea_year cluster_zd_year cluster_class_year inventor cluster1 cluster_bea_class cluster_class_year org_new) vce(cluster cluster1) keepsin 
+
+
+*****************
+* 2 MONTHS
+*****************
+clear
+u $main/data2/data_4e
+egen total = sum(number), by(inventor)
+
+g y = log(number)
+g x   = log(Den_bea_zd    )
+
+egen cluster1               = group(bea zd)
+egen cluster_bea_year       = group(bea year)
+egen cluster_bea_zd_year    = group(bea zd year)
+egen cluster_zd_year        = group(zd year)
+egen cluster_class_year     = group(class year)
+egen cluster_bea_class      = group(bea class)
+egen org_new2               = group(org_new)
+drop org_new
+
+keep if total  >=3.25
+keep if y~=. & zd ~=. & class ~=. & year ~=. & inventor ~=. & cluster1 ~=. & bea ~=.
+summ y x
+
+eststo: reghdfe y x , absorb(year bea zd class cluster_bea_year cluster_zd_year cluster_class_year inventor cluster1 cluster_bea_class cluster_class_year org_new) vce(cluster cluster1)  keepsin 
+
+*****************
+* 3 MONTH
+*****************
+clear
+u $main/data2/data_4f
+egen total = sum(number), by(inventor)
+
+g y = log(number)
+g x   = log(Den_bea_zd    )
+
+egen cluster1               = group(bea zd)
+egen cluster_bea_year       = group(bea year)
+egen cluster_bea_zd_year    = group(bea zd year)
+egen cluster_zd_year        = group(zd year)
+egen cluster_class_year     = group(class year)
+egen cluster_bea_class      = group(bea class)
+egen org_new2               = group(org_new)
+drop org_new
+
+keep if total  >=3.25
+keep if y~=. & zd ~=. & class ~=. & year ~=. & inventor ~=. & cluster1 ~=. & bea ~=.
+summ y x
+
+eststo: reghdfe y x , absorb(year bea zd class cluster_bea_year cluster_zd_year cluster_class_year inventor cluster1 cluster_bea_class cluster_class_year org_new) vce(cluster cluster1)  keepsin 
+
+
+*****************
+* 6 MONTH
+*****************
+clear
+u $main/data2/data_4b
+egen total = sum(number), by(inventor)
+
+g y = log(number)
+g x   = log(Den_bea_zd    )
+
+egen cluster1               = group(bea zd)
+egen cluster_bea_year       = group(bea year)
+egen cluster_bea_zd_year    = group(bea zd year)
+egen cluster_zd_year        = group(zd year)
+egen cluster_class_year     = group(class year)
+egen cluster_bea_class      = group(bea class)
+egen org_new2               = group(org_new)
+drop org_new
+
+
+keep if total  >=3.25
+keep if y~=. & zd ~=. & class ~=. & year ~=. & inventor ~=. & cluster1 ~=. & bea ~=.
+summ y x 
+
+eststo: reghdfe y x , absorb(year bea zd class cluster_bea_year cluster_zd_year cluster_class_year inventor cluster1 cluster_bea_class cluster_class_year org_new) vce(cluster cluster1) keepsin 
+
+*****************
+* 1 YEAR
+*****************
+clear
+u $main/data2/data_3
+egen total = sum(number), by(inventor)
+
+g y = log(number)
+g x   = log(Den_bea_zd    )
+
+egen cluster1               = group(bea zd)
+egen cluster_bea_year       = group(bea year)
+egen cluster_bea_zd_year    = group(bea zd year)
+egen cluster_zd_year        = group(zd year)
+egen cluster_class_year     = group(class year)
+egen cluster_bea_class      = group(bea class)
+egen org_new                = group(org_id)
+
+keep if total  >=3.25
+keep if y~=. & zd ~=. & class ~=. & year ~=. & inventor ~=. & cluster1 ~=. & bea ~=.
+summ y x 
+
+eststo: reghdfe y x , absorb(year bea zd class cluster_bea_year cluster_zd_year cluster_class_year inventor cluster1 cluster_bea_class cluster_class_year org_new) vce(cluster cluster1) keepsin 
+
+*****************
+* 2 YEAR
+*****************
+clear
+u $main/data2/data_4c
+egen total = sum(number), by(inventor)
+
+g y = log(number)
+g x   = log(Den_bea_zd    )
+
+egen cluster1               = group(bea zd)
+egen cluster_bea_year       = group(bea year)
+egen cluster_bea_zd_year    = group(bea zd year)
+egen cluster_zd_year        = group(zd year)
+egen cluster_class_year     = group(class year)
+egen cluster_bea_class      = group(bea class)
+egen org_new2               = group(org_new)
+drop org_new
+
+keep if total  >=3.25
+keep if y~=. & zd ~=. & class ~=. & year ~=. & inventor ~=. & cluster1 ~=. & bea ~=.
+summ y x 
+
+eststo: reghdfe y x , absorb(year bea zd class cluster_bea_year cluster_zd_year cluster_class_year inventor cluster1 cluster_bea_class cluster_class_year org_new) vce(cluster cluster1) keepsin 
+
+*****************
+* 3 YEAR
+*****************
+clear
+u $main/data2/data_4d
+egen total = sum(number), by(inventor)
+
+g y = log(number)
+g x   = log(Den_bea_zd    )
+
+egen cluster1               = group(bea zd)
+egen cluster_bea_year       = group(bea year)
+egen cluster_bea_zd_year    = group(bea zd year)
+egen cluster_zd_year        = group(zd year)
+egen cluster_class_year     = group(class year)
+egen cluster_bea_class      = group(bea class)
+egen org_new2               = group(org_new)
+drop org_new
+
+
+keep if total  >=3.25
+keep if y~=. & zd ~=. & class ~=. & year ~=. & inventor ~=. & cluster1 ~=. & bea ~=.
+summ y x 
+
+eststo: reghdfe y x , absorb(year bea zd class cluster_bea_year cluster_zd_year cluster_class_year inventor cluster1 cluster_bea_class cluster_class_year org_new) vce(cluster cluster1) keepsin 
+esttab using tables4/table9.tex, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+esttab using tables4/table9.txt, se  compress replace star(* 0.10 ** 0.05 *** 0.01)
+eststo clear
+
+
+restore
+

107/replication_package/AER_UPLOADED/reg5.do

@@ -0,0 +1,88 @@
+********************************
+* Firm specific spillovers
+********************************
+
+
+u $main/data2/data_3_count
+keep if year ==2007
+keep bea zd2 year count total
+summ
+
+sort bea zd 
+collapse count total, by(bea zd)
+save tmp, replace
+summ
+
+
+**********************************************************
+u $main/data2/data_patent_level_3
+drop if org_id ==""
+keep if year ==2007
+
+sort inventor_id bea zd
+by inventor_id  bea zd: g d1 = 1 if _n ==1
+
+g org_N = d1 
+sort org_norm org_id zd2 bea
+collapse (sum) org_N, by(org_norm org_id zd2 bea)
+summ
+
+* ADD COUNT data
+sort bea zd
+merge m:1 bea zd using tmp
+tab _merge
+drop _merge
+
+
+* IMPACT OF FIRM
+g share1 = org_N / count
+g D_prod = 0.066*( log(count/total) -  log((count-org_N)/total)  )
+
+summ share1 D_prod if zd ==2
+summ share1 D_prod if zd ==1
+summ share1 D_prod if zd ==5
+
+
+* COMPUTER SCIENCE
+* XXX SPILLOVERES
+preserve
+keep if zd2 ==2
+gsort -org_N
+g n = _n
+keep if n <= 100
+drop zd n
+list
+
+
+* BIOLOGY CHEMETYRY
+* XXX SPILLOVERS
+restore
+preserve
+keep if zd2 ==1
+gsort -org_N
+g n = _n
+keep if n <= 100
+drop zd n
+list
+
+
+* SEMICONDUCTORS
+restore
+preserve
+keep if zd2 ==5
+gsort -org_N
+g n = _n
+keep if n <= 100
+drop zd n
+list
+
+
+
+
+
+
+
+
+
+
+

107/replication_package/README.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1c7c5f5dcd8cf6a025631b08e8423ae45c201ccff006098d64b1b970c3e43fdd
+size 354330

107/should_reproduce.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eccda9d44811a8d1afe36a59bbe6d69d50fda658b88215d8e0b8accc7753b1fc
+size 16