* ############
* # READ-ME  #
* ############
* This STATA 11 do-file  replicates findings in 
* "The State of the Labor Market for Registered Nurses."
* To begin, you must download the CPS-MORG files from 2000 
* to 2009, available online at 
* http://www.nber.org/morg/annual/
* and migrate them to your STATA working directory. 
* For convenience, a file "nursemorgs.dta" is created, from 
* which any latter analysis may be easily performed. 
*
* This do-file was authored by Alan Benson (alanmb@mit.edu), 
* on June 21-30, 2010. 

cd "WORKING DIRECTORY NAME HERE"

clear all
set mem 1500m

use morg00.dta
for num 1/9: append using morg0X.dta
set scheme s2mono

egen popx3=total(weight), by(year)
gen pop=popx3/3
gen cohort=year-age
egen popx3_ageco=total(weight), by(age cohort)
gen pop_ageco=popx3_ageco/3
drop popx3 popx3_ageco

label define lbl_lfsr94 1 "E: Working" 2 "E: With Job, Not at Work" 3 "U: Layoff" 4 "U: Looking" 5 "Other" 6 "Disabled" 7 "Other"
label values lfsr94 lbl_lfsr94
keep if lfsr94<=4
gen one_all=1

* Identify RNs
gen rn=0
replace rn=1 if occ00==3130 & year>=2000
label define lbl_rn 0 "Not Nurse" 1 "RN"
label value rn lbl_rn

* Identify Healthcare Occupations from 2000 SOC Codes Major Groups 29 & 31
gen health=0
replace health=1 if year>=2000 & (occ00>=3000 & occ00<=3650)
label var health "0 Not Health-Care Occ, 1 Health Care Occ"
label define lbl_health 0 "Non-Healthcare Occ" 1 "Healthcare Occ"
label value health lbl_health

* Generate Counts by Group
gen one_rn=0
replace one_rn=1 if rn==1
gen one_health=0
replace one_health=1 if health==1

* Center Weights (Mean of One)
egen temp1=mean(weight) if lfsr>=1 & rn==1, by(year) 
gen wt_rn=weight/temp1 if lfsr>=1 & rn==1
egen temp2=mean(weight) if lfsr>=1 & health==1, by(year) 
gen wt_health=weight/temp2 if lfsr>=1 & health==1
egen temp3=mean(weight) if lfsr>=1, by(year) 
gen wt_all=weight/temp3 if lfsr>=1
for num 1/3: drop tempX

save "nursemorgs.dta", replace

* Table 1(i): Unemployment rates
preserve
gen ue_rn=.
replace ue_rn=0 if lfsr94<=2 & rn==1
replace ue_rn=wt_rn if lfsr94>=3 & rn==1
gen ue_health=.
replace ue_health=0 if lfsr94<=2 & health==1
replace ue_health=wt_health if lfsr94>=3 & health==1
gen ue_all=.
replace ue_all=0 if lfsr94<=2
replace ue_all=wt_all if lfsr94>=3 
collapse (mean) ue_rn ue_health ue_all (sum) one_rn one_health one_all, by(year)
gen hi_rn=ue_rn+2*sqrt((ue_rn*(1-ue_rn))/one_rn)
gen lo_rn=ue_rn-2*sqrt((ue_rn*(1-ue_rn))/one_rn)
gen hi_health=ue_health+2*sqrt((ue_health*(1-ue_health))/one_health)
gen lo_health=ue_health-2*sqrt((ue_health*(1-ue_health))/one_health)
gen hi_all=ue_all+2*sqrt((ue_all*(1-ue_all))/one_all)
gen lo_all=ue_all-2*sqrt((ue_all*(1-ue_all))/one_all)
# delimit ;
scatter ue_rn year, pstyle (p2)|| 
scatter ue_health year, pstyle (p2) ||
scatter ue_all year, pstyle (p2) ||
line ue_rn year, pstyle (p2) sort || 
line ue_health year, pstyle (p3) sort ||
line ue_all year, pstyle (p4) sort
;
# delimit cr
restore 

* Table 1(ii) FTEs Per Capita
preserve
gen fte_rn=((52*uhours*weight)/3)/2000 if rn==1
gen fte_health=((52*uhours*weight)/3)/2000 if health==1
gen fte_all=((52*uhours*weight)/3)/2000
collapse (sum) fte_rn fte_health fte_all wt_rn one_all (median) pop, by(year)
rename fte_rn fterns
rename fte_health ftehealth
rename fte_all fteall
gen ratio_rn=fterns/pop
gen ratio_health=ftehealth/pop
gen ratio_all=fteall/pop
gen hi_rn=ratio_rn+2*sqrt((ratio_rn*(1-ratio_rn))/one_all)
gen lo_rn=ratio_rn-2*sqrt((ratio_rn*(1-ratio_rn))/one_all)
scatter ratio_rn year, pstyle(p1 p1 p1) sort || line ratio_rn hi_rn lo_rn year, pstyle(p1) lpattern(_ .) sort
restore

* Table 1(iii): LT 35 Hrs due to Slack Work
preserve 
gen slack_rn=.
replace slack_rn=0 if rn==1
replace slack_rn=wt_rn if reason94==1 & rn==1
gen slack_health=.
replace slack_health=0 if health==1
replace slack_health=wt_health if reason94==1 & health==1
gen slack_all=0
replace slack_all=wt_all if reason94==1
collapse (mean) slack_rn slack_health slack_all (sum) one_rn one_health one_all, by(year)
gen hi_rn=slack_rn+2*sqrt((slack_rn*(1-slack_rn))/one_rn)
gen lo_rn=slack_rn-2*sqrt((slack_rn*(1-slack_rn))/one_rn)
gen hi_health=slack_health+2*sqrt((slack_health*(1-slack_health))/one_health)
gen lo_health=slack_health-2*sqrt((slack_health*(1-slack_health))/one_health)
gen hi_all=slack_all+2*sqrt((slack_all*(1-slack_all))/one_all)
gen lo_all=slack_all-2*sqrt((slack_all*(1-slack_all))/one_all)
*# delimit ;
*scatter slack_rn year, pstyle (p2)|| 
*scatter slack_health year, pstyle (p2)|| 
*scatter slack_all year, pstyle (p2)|| 
*line slack_rn lo_rn hi_rn year, pstyle (p2 p2 p2) sort ||
*line slack_health lo_health hi_health year, pstyle (p3 p3 p3) sort ||
*line slack_all lo_all hi_all year, pstyle (p4 p4 p4) sort 
*;
*# delimit cr
scatter slack_rn slack_health slack_all year, pstyle(p2 p3 p4) sort|| line slack_rn slack_health slack_all year, pstyle (p2 p3 p4) lpattern(longdash shortdash solid) sort

restore 

* Table 2: Reason LT 35 Hrs/Week
preserve
label define lbl_reason94 1 "Slack Work" 2 "Seasonal" 3 "New Job Dur Wk" 4 "On Vacation" 5 "Illness" 6 "Holiday" 7 "Child Care" 8 "Other Family" 9 "Labor Dispute" 10 "Weather" 11 "School/Training" 12 "Civic/Mil Duty" 13 "Other"
label values reason94 lbl_reason94
gen whylt35=.
replace whylt35=1 if reason94==1
replace whylt35=2 if reason94==4
replace whylt35=3 if reason94==6
replace whylt35=4 if reason94==5
replace whylt35=5 if reason94==2 | reason94==3 | reason94>7
replace whylt35=. if reason94==.
label define lbl_whylt35 1 "Slack Work" 2 "Vacation" 3 "Holiday" 4 "Illness" 5 "Other"
label values whylt35 lbl_whylt35
gen period=0 
replace period=1 if year>=2005 & year<=2008
replace period=2 if year==2009
label define lbl_period 0 "2000-2004" 1 "2005-2008" 2 "2009"
label values period lbl_period
gen temp=round(weight,1)
tab2 whylt35 period [fw=temp], col nofreq
restore

* TEXT: Claim regarding young workers
preserve
keep if age>=20 & age<=29
gen ue_rn=.
replace ue_rn=0 if lfsr94<=2 & rn==1
replace ue_rn=wt_rn if lfsr94>=3 & rn==1
gen ue_health=.
replace ue_health=0 if lfsr94<=2 & health==1
replace ue_health=wt_health if lfsr94>=3 & health==1
gen ue_all=.
replace ue_all=0 if lfsr94<=2
replace ue_all=wt_all if lfsr94>=3 
collapse (mean) ue_rn ue_health ue_all (sum) one_rn one_health one_all, by(year)
tab year, sum(ue_rn) mean
tab year, sum(ue_all) mean
gen hi_rn=ue_rn+2*sqrt((ue_rn*(1-ue_rn))/one_rn)
gen lo_rn=ue_rn-2*sqrt((ue_rn*(1-ue_rn))/one_rn)
gen hi_health=ue_health+2*sqrt((ue_health*(1-ue_health))/one_health)
gen lo_health=ue_health-2*sqrt((ue_health*(1-ue_health))/one_health)
gen hi_all=ue_all+2*sqrt((ue_all*(1-ue_all))/one_all)
gen lo_all=ue_all-2*sqrt((ue_all*(1-ue_all))/one_all)
# delimit ;
scatter ue_rn year, pstyle (p2)|| 
scatter ue_health year, pstyle (p2) ||
scatter ue_all year, pstyle (p2) ||
line ue_rn year, pstyle (p2) sort || 
line ue_health year, pstyle (p3) sort ||
line ue_all year, pstyle (p4) sort
;
# delimit cr
restore

* TEXT: On Layoff
preserve
gen lay_rn=.
replace lay_rn=0 if lfsr94<=2 & rn==1
replace lay_rn=wt_rn if lfsr94==3 & rn==1
gen lay_health=.
replace lay_health=0 if lfsr94<=2 & health==1
replace lay_health=wt_health if lfsr94==3 & health==1
gen lay_all=.
replace lay_all=0 if lfsr94<=2
replace lay_all=wt_all if lfsr94==3 
sum lay_rn if year==2009
sum lay_rn if year<2009
sum lay_all if year==2009
sum lay_all if year<2009
restore 

* Table 3: PT/FT Categories
preserve
replace uhours=. if uhours==99
gen hourscat=.
replace hourscat=1 if uhourse>=0  & uhourse<=19
replace hourscat=2 if uhourse>=20 & uhourse<=29
replace hourscat=3 if uhourse>30  & uhourse<=34
replace hourscat=4 if uhourse>35  & uhourse<=39
replace hourscat=5 if uhourse==40
replace hourscat=6 if uhourse>=41 & uhourse<=49
replace hourscat=7 if uhourse>50  & uhourse<=59
replace hourscat=8 if uhourse>=60
gen period=1 
replace period=2 if year>=2005 & year<=2008
replace period=3 if year==2009
label define lbl_period 1 "2000-2004" 2 "2005-2008" 3 "2009"
label values period lbl_period
label define lbl_hourscat 1 "0-19" 2 "20-29" 3 "30-34" 4 "35-39" 5 "40" 6 "41-49" 7 "50-59" 8 "60+"
label values hourscat lbl_hourscat
gen temp=round(weight,1)
tab2 hourscat period [fw=temp], col nofreq
restore

* Table 4(i): Median Age
preserve
gen age_rn=age if rn==1
gen age_health=age if health==1
gen age_all=age
collapse (median) p50_age_rn=age_rn p50_age_health=age_health p50_age_all=age_all (p25) p25_age_rn=age_rn p25_age_health=age_health p25_age_all=age_all (p75) p75_age_rn=age_rn p75_age_health=age_health p75_age_all=age_all (sum) one_rn one_health one_all, by(year)
# delimit ;
scatter p50_age_rn year, pstyle (p2)|| 
scatter p50_age_health year, pstyle (p2) ||
scatter p50_age_all year, pstyle (p2) ||
line p50_age_rn year, pstyle (p1 p1 p1) sort || 
line p50_age_health year, pstyle (p3 p3 p3) sort ||
line  p50_age_al year, pstyle (p4 p4 p4) sort
;
# delimit cr
restore 

* Table 4(ii): Age distribution change in RNs from 2008 to 2009
preserve
gen age_rn=age if rn==1
gen age_health=age if health==1
gen age_all=age
keep if year==2009 | year==2000
collapse (sum) wt_rn, by(age year)
gen pdf_rn=wt_rn/4068 if year==2009
replace pdf_rn=wt_rn/3040 if year==2000
line pdf_rn age if year==2000, pstyle(p2) sort || line pdf_rn age if year==2009, pstyle(p3) sort
restore

* Table 5: Demographics
preserve
gen male=2-sex
gen noncit=.
replace noncit=0 if prcitshp>=1 & prcitshp<=4 
replace noncit=1*wt_all if prcitshp==5 
gen nonwhite=.
replace nonwhite=wt_all if race>1
replace nonwhite=0 if race==1
*gen married=.
*replace married=1 if marital<4
*replace married=0 if marital>=4
gen rn_male=2-sex if rn==1
gen rn_noncit=.
replace rn_noncit=0 if prcitshp>=1 & prcitshp<=4 & rn==1
replace rn_noncit=wt_rn if prcitshp==5 & rn==1
gen rn_nonwhite=.
replace rn_nonwhite=wt_rn if race>1 & rn==1
replace rn_nonwhite=0 if race==1 & rn==1
gen rn_young=.
replace rn_young=wt_rn if age<=34 & rn==1
replace rn_young=0 if age>=35 & rn==1
gen rn_forborn=.
replace rn_forborn=wt_rn if penat!=57 & rn==1
replace rn_forborn=0 if penat==57 & rn==1
*gen rn_married=.
*replace rn_married=1 if marital<4 & rn==1
*replace rn_married=0 if marital>=4 & rn==1
collapse (mean) male noncit nonwhite rn_male rn_noncit rn_nonwhite rn_young rn_forborn, by(year)
# delimit ;
scatter rn_male year, pstyle(p1) ||
scatter rn_noncit year, pstyle(p1) ||
scatter rn_young year, pstyle(p1) ||
scatter rn_forborn year, pstyle(p1) ||
scatter rn_nonwhite year, pstyle(p1) ||
line rn_male year, lpattern(l) pstyle(p1) ||
line rn_noncit year, lpattern(l) pstyle(p2) ||
line rn_young year, lpattern(l) pstyle(p3) ||
line rn_forborn year, lpattern(l) pstyle(p4) ||
line rn_nonwhite year, lpattern(l) pstyle(p5) 
;
# delimit cr
restore

* Table 7: Compensation Profiles
preserve
gen age_xrn=age*rn
gen agesq=age^2
gen agesq_xrn=agesq*rn
gen female=sex-1
gen female_xrn=female*rn
gen realwage=earnhre
replace realwage=earnhre/100 if year==2009
replace realwage=(0.99600*earnhre)/100 if year==2008
replace realwage=(1.03385*earnhre)/100 if year==2007
replace realwage=(1.06280*earnhre)/100 if year==2006
replace realwage=(1.08937*earnhre)/100 if year==2005
replace realwage=(1.12640*earnhre)/100 if year==2004
replace realwage=(1.16358*earnhre)/100 if year==2003
replace realwage=(1.18568*earnhre)/100 if year==2002
replace realwage=(1.21414*earnhre)/100 if year==2001
replace realwage=(1.23357*earnhre)/100 if year==2000
replace realwage=. if earnhre==0 | earnhre==9999
gen rw_malernasso=realwage if female==0 & rn==1 & (ihigrdc==14 | ihigrdc==15)
gen rw_malernbach=realwage if female==0 & rn==1 & (ihigrdc==16)
gen rw_malerngrad=realwage if female==0 & rn==1 & (ihigrdc>16)
gen rw_maleallasso=realwage if female==0 & (ihigrdc==14 | ihigrdc==15)
gen rw_maleallbach=realwage if female==0 & (ihigrdc==16)
gen rw_maleallgrad=realwage if female==0 & (ihigrdc>16)
gen rw_femrnasso=realwage if female==1 & rn==1 & (ihigrdc==14 | ihigrdc==15)
gen rw_femrnbach=realwage if female==1 & rn==1 & (ihigrdc==16)
gen rw_femrngrad=realwage if female==1 & rn==1 & (ihigrdc>16)
gen rw_femallasso=realwage if female==1 & (ihigrdc==14 | ihigrdc==15)
gen rw_femallbach=realwage if female==1 & (ihigrdc==16)
gen rw_femallgrad=realwage if female==1 & (ihigrdc>16)
qreg rw_malernasso age agesq, nolog
predict yhat_malernasso if female==0 & rn==1 & (ihigrdc==14 | ihigrdc==15)
qreg rw_malernbach age agesq, nolog
predict yhat_malernbach if female==0 & rn==1 & (ihigrdc==16)
qreg rw_malerngrad age agesq, nolog
predict yhat_malerngrad if female==0 & rn==1 & (ihigrdc>16)
qreg rw_maleallasso age agesq, nolog
predict yhat_maleallasso if female==0 & (ihigrdc==14 | ihigrdc==15)
qreg rw_maleallbach age agesq, nolog
predict yhat_maleallbach if female==0 & (ihigrdc==16)
qreg rw_maleallgrad age agesq, nolog
predict yhat_maleallgrad if female==0 & (ihigrdc>16)
qreg rw_femrnasso age agesq, nolog
predict yhat_femrnasso if female==1 & rn==1 & (ihigrdc==14 | ihigrdc==15)
qreg rw_femrnbach age agesq, nolog
predict yhat_femrnbach if female==1 & rn==1 & (ihigrdc==16)
qreg rw_femrngrad age agesq, nolog
predict yhat_femrngrad if female==1 & rn==1 & (ihigrdc>16)
qreg rw_femallasso age agesq, nolog
predict yhat_femallasso if female==1 & (ihigrdc==14 | ihigrdc==15)
qreg rw_femallbach age agesq, nolog
predict yhat_femallbach if female==1 & (ihigrdc==16)
qreg rw_femallgrad age agesq, nolog
predict yhat_femallgrad if female==1 & (ihigrdc>16)
*CHOW TESTS
gen femxage=female*age
gen femxagesq=female*agesq
qreg realwage female age agesq femxage femxagesq if (ihigrdc==14 | ihigrdc==15) & rn==1, nolog
test femxage=femxagesq=0
qreg realwage female age agesq femxage femxagesq if (ihigrdc==16) & rn==1, nolog
test femxage=femxagesq=0
qreg realwage female age agesq femxage femxagesq if (ihigrdc>16) & rn==1, nolog
test femxage=femxagesq=0
collapse (median) rw_malernasso rw_malernbach rw_malerngrad rw_maleallasso rw_maleallbach rw_maleallgrad rw_femrnasso rw_femrnbach rw_femrngrad rw_femallasso rw_femallbach rw_femallgrad  yhat_malernasso yhat_malernbach yhat_malerngrad yhat_maleallasso yhat_maleallbach yhat_maleallgrad yhat_femrnasso yhat_femrnbach yhat_femrngrad yhat_femallasso yhat_femallbach yhat_femallgrad, by(age)
scatter rw_femrnasso rw_malernasso rw_femallasso rw_maleallasso age if age<=70, pstyle(p1 p1 p1 p1) msymbol(o oh o oh) sort || line yhat_femrnasso yhat_malernasso yhat_femallasso yhat_maleallasso age if age<=70, pstyle(p1 p1 p1 p1) sort
scatter rw_femrnbach rw_malernbach rw_femallbach rw_maleallbach age if age<=70 & rw_malernbach<50, pstyle(p1 p1 p1 p1) msymbol(o oh o oh) sort || line yhat_femrnbach yhat_malernbach yhat_femallbach yhat_maleallbach age if age<=70, pstyle(p1 p1 p1 p1) sort
scatter rw_femrngrad rw_malerngrad rw_femallgrad rw_maleallgrad age if age<=70, pstyle(p1 p1 p1 p1) msymbol(o oh o oh) sort || line yhat_femrngrad yhat_malerngrad yhat_femallgrad yhat_maleallgrad age if age<=70, pstyle(p1 p1 p1 p1) sort
keep if (age==25 | age==35 | age==45 | age==55)
gen prem_maleasso=yhat_malernasso/yhat_maleallasso
gen prem_malebach=yhat_malernbach/yhat_maleallbach
gen prem_malegrad=yhat_malerngrad/yhat_maleallgrad
gen prem_maleasso=yhat_malernasso/yhat_maleallasso
gen prem_malebach=yhat_malernbach/yhat_maleallbach
gen prem_malegrad=yhat_malerngrad/yhat_maleallgrad
browse
restore

* Table 8: Compensation Tabulation
preserve
gen age_xrn=age*rn
gen agesq=age^2
gen agesq_xrn=agesq*rn
gen female=sex-1
gen female_xrn=female*rn
gen realwage=earnhre
replace realwage=earnhre/100 if year==2009
replace realwage=(0.99600*earnhre)/100 if year==2008
replace realwage=(1.03385*earnhre)/100 if year==2007
replace realwage=(1.06280*earnhre)/100 if year==2006
replace realwage=(1.08937*earnhre)/100 if year==2005
replace realwage=(1.12640*earnhre)/100 if year==2004
replace realwage=(1.16358*earnhre)/100 if year==2003
replace realwage=(1.18568*earnhre)/100 if year==2002
replace realwage=(1.21414*earnhre)/100 if year==2001
replace realwage=(1.23357*earnhre)/100 if year==2000
replace realwage=. if earnhre==0 | earnhre==9999
gen rw_malernasso=realwage if female==0 & rn==1 & (ihigrdc==14 | ihigrdc==15)
gen rw_malernbach=realwage if female==0 & rn==1 & (ihigrdc==16)
gen rw_malerngrad=realwage if female==0 & rn==1 & (ihigrdc>16)
gen rw_maleallasso=realwage if female==0 & (ihigrdc==14 | ihigrdc==15)
gen rw_maleallbach=realwage if female==0 & (ihigrdc==16)
gen rw_maleallgrad=realwage if female==0 & (ihigrdc>16)
gen rw_femrnasso=realwage if female==1 & rn==1 & (ihigrdc==14 | ihigrdc==15)
gen rw_femrnbach=realwage if female==1 & rn==1 & (ihigrdc==16)
gen rw_femrngrad=realwage if female==1 & rn==1 & (ihigrdc>16)
gen rw_femallasso=realwage if female==1 & (ihigrdc==14 | ihigrdc==15)
gen rw_femallbach=realwage if female==1 & (ihigrdc==16)
gen rw_femallgrad=realwage if female==1 & (ihigrdc>16)
*CALCULATE GREATEST SEs
qreg rw_malernasso age agesq, nolog
lincom 35*age+1225*agesq+_cons
lincom 45*age+2025*agesq+_cons
lincom 55*age+3025*agesq+_cons
qreg rw_maleallasso age agesq, nolog
lincom 35*age+1225*agesq+_cons
lincom 45*age+2025*agesq+_cons
lincom 55*age+3025*agesq+_cons
qreg rw_malernbach age agesq, nolog
lincom 35*age+1225*agesq+_cons
lincom 45*age+2025*agesq+_cons
lincom 55*age+3025*agesq+_cons
qreg rw_maleallbach age agesq, nolog
lincom 35*age+1225*agesq+_cons
lincom 45*age+2025*agesq+_cons
lincom 55*age+3025*agesq+_cons
qreg rw_malerngrad age agesq, nolog
lincom 35*age+1225*agesq+_cons
lincom 45*age+2025*agesq+_cons
lincom 55*age+3025*agesq+_cons
qreg rw_maleallgrad age agesq, nolog
lincom 35*age+1225*agesq+_cons
lincom 45*age+2025*agesq+_cons
lincom 55*age+3025*agesq+_cons
qreg rw_femrnasso age agesq, nolog
lincom 35*age+1225*agesq+_cons
lincom 45*age+2025*agesq+_cons
lincom 55*age+3025*agesq+_cons
qreg rw_femallasso age agesq, nolog
lincom 35*age+1225*agesq+_cons
lincom 45*age+2025*agesq+_cons
lincom 55*age+3025*agesq+_cons
qreg rw_femrnbach age agesq, nolog
lincom 35*age+1225*agesq+_cons
lincom 45*age+2025*agesq+_cons
lincom 55*age+3025*agesq+_cons
qreg rw_femallbach age agesq, nolog
lincom 35*age+1225*agesq+_cons
lincom 45*age+2025*agesq+_cons
lincom 55*age+3025*agesq+_cons
qreg rw_femrngrad age agesq, nolog
lincom 35*age+1225*agesq+_cons
lincom 45*age+2025*agesq+_cons
lincom 55*age+3025*agesq+_cons
qreg rw_femallgrad age agesq, nolog
lincom 35*age+1225*agesq+_cons
lincom 45*age+2025*agesq+_cons
lincom 55*age+3025*agesq+_cons

* SECTION II: Forecast 
preserve
replace uhours=. if uhours==99
*age 25-75 at some point in sample
keep if cohort>=1925 & cohort<=1985
egen rnsx3=total(weight) if rn==1, by(age cohort)
gen rnhours=(rnsx3*uhours*52)/3
gen rnftes=rnhours/2000
gen weight_rn=weight if rn==1
collapse (median) rnftes pop_ageco (sum) weight_rn, by(age cohort)
gen share=rnftes/pop_ageco
gen lnshare=log(rnftes/pop_ageco)
anova lnshare cohort age [aw=weight_rn]
regress
* Steps for reproduction are as follows:
* Population data is taken from 2020 Census "Middle-Range Projections." 
* Cohort is calculated by 2010-AGE. 
* Share of FTE RNs at each age are estimated using the model. 
* This share is then multiplied by projected population. 
* This column is now the projected number of RNs at each age. 
* This column is summed to reach an estimated total number of RNs. 
* Column is then divided by total estimated number of RNs. 
* This column is now the "PDF" graphed in Table 10.
restore 

* Table 9: Cohort and Age Effects
preserve
clear all
use forecast.dta
rename v1 cohort
rename v2 cohort_coeff
rename v3 age
rename v4 age_coeff
reg cohort_coeff cohort if cohort>=1965
predict cocohat if cohort>=1965
line cohort_coeff cohort, lstyle(p1) || scatter cocohat cohort if cohort>=1985, msymbol(oh) || line cocohat cohort if cohort>=1965, lstyle(.)
line age_coeff age, lstyle(p1)
restore

* Table 10: RN Age Profile Projection
preserve
gen age_rn=age if rn==1
gen age_health=age if health==1
gen age_all=age
keep if year==2009 | year==2000
collapse (sum) wt_rn, by(age year)
gen pdf_rn=wt_rn/4068 if year==2009
replace pdf_rn=wt_rn/3040 if year==2000
append using  "/mnt/san0/alanmb/94 Nurse Inst/Forecast/forecast20002020.dta"
line pdf_rn age if year==2000, pstyle(p1) lpattern(dot) sort || line pdf_rn age if year==2009, pstyle(p3) lpattern(_) sort || line pdf_rn age if year==2020 & age<=75, pstyle(p4) lpattern(solid) sort
restore