The following files (all appended below) should be run in LISSY, in the order provided:

Transcription

1 REPLICATION PACKAGE INSTRUCTIONS Brady, David, Ryan Finnigan, and Sabine Huebgen Rethinking the Risks of Poverty: A Framework for Analyzing Prevalences and Penalties. Forthcoming at the American Journal of Sociology. This file outlines the Stata do files contained in this replication package, and their correspondence to the tables and figures in the article. The files were written and revised mostly by Sabine Huebgen throughout , based on much more limited files written by David Brady and Ryan Finnigan in This package was assembled by Ryan Finnigan (rfinnigan@ucdavis.edu) in January DATA DESCRIPTION The primary data for the paper are available from the Luxembourg Income Study (LIS). The LIS is accessible with registration through LIS registration is free for students, and fee requirements are posted at The LIS data may not be downloaded, and must be accessed remotely through the LISSY interface ( Do files can be submitted to LISSY as jobs, and results are returned as text only. The replication package splits the do files between those run in the LISSY interface and those run locally based on the LISSY results. DESCRIPTION OF REPLICATION FILES The following files (all appended below) should be run in LISSY, in the order provided: BFH_LIS_1_data.do: pools together data on the 29 countries in the sample, and generates all key variables BFH_LIS_2_figs1to3.do: produces the underlying estimates for figures 1-3. The prevalence and penalty estimates are also saved in Stata format for use in BFH_local_1_figs1to4.do and BFH_local_2_simulations.do BFH_LIS_3_figs6_7.do: produces the underlying estimates for figures 6 & 7, using median prevalence/penalty estimates from BFH_local_2_simulations.do BFH_LIS_4_tables2_3.do: produces the regression tables 2 and 3 BFH_LIS_5_appendix.do: produces the estimates for all the tables and figures in the appendix The results from these do files were copied into the excel file BFH_LISSY_results.xls. These results were then imported into Stata locally to produce the article s figures and tables in the file BFH_local.do.

2 BFH_LIS_1_data.do * Replication file for Brady, Finnigan, and Huebgen. "Rethinking the Risks of Poverty: A Framework for Analyzing Prevalences and Penalties." /* This do file contains the Stata code for constructing the cross-national data file for the article. The following code can be submitted to the LISSY interface of the Luxembourg Income Study ( by registered users. Please see the instructions (BFH_replication_instructions.txt) in this package for more information. The original version of the code in this file was written by Sabine Huebgen, Nov 10, Some edits by Ryan Finnigan, January */ // 29 countries: HU05 included, TW dropped *** Loop for create the country files and put them together *** except Switzerland --> nearn has to be calculated by hand *** except Hungary --> educ has to be recoded global c "au10 at04 be00 ca10 cz10 dk10 ee10 fi10 fr10 de10 gr10 ie10 il10 it10 is10 jp08 lu10 nl10 no10 pl10 sk10 si10 kr06 es10 se05 uk10 us13" foreach x of global c { *HH file use $`x'h, clear drop if dhi==. drop if dhi==0 drop if hwgt==. gen ey=dhi/(sqrt(nhhmem)) qui sum ey gen botlin=0.01*_result(3) replace ey=botlin if ey<botlin quietly sum ey, de gen toplin=10*_result(10) gen multearn=. replace multearn=0 if nearn==0 nearn==1 replace multearn=1 if nearn>1 & nearn!=. gen unemphh=. replace unemphh=0 if nearn>0 & nearn!=. replace unemphh=1 if nearn==0 sort hid keep did year hid hwgt nhhmem hhtype nhhmem65 nhhmem17 ey hi hil hile hic hicid hit dhi hpartner hitp hits hitsi hitsu hitsa nearn unemphh multearn save $mydata/ `x'h, replace *Person File

3 use $`x'p, clear gen head=. replace head=1 if relation==1000 replace head=0 if relation>1000 & relation!=. recode sex (1=0)(2=1)(.=.), gen(female) recode sex (1=1)(2=0)(.=.), gen(male) sort hid keep hid pid did year hhmem relation partner children nchildren ageyoch age sex immigr yrsresid ethnic_c educ emp hours secjob inda1 indb1 indc1 sector1 occa1 occb1 pmi pmxit pmxiti ppension pi pil pmil pi pile pmile pils pmils pit pmit pits pmits pitsis pitsisun pmitsisun ptime gross1 educ_c marital head male female save $mydata/`x'p, replace merge m:1 hid using $mydata/brady/`x'h, keepusing (did year hid hwgt nhhmem hhtype nhhmem65 nhhmem17 ey hi hil hile hic hicid hit dhi hpartner hitp hits hitsi hitsu hitsa nearn unemphh multearn) keep(match) nogen *create variable for lead earner* recode pil (.=0) egen maxinc=max(pil), by(hid) gen lead=pil==maxinc egen maxage=max(age) if lead, by(hid) replace lead=0 if age~=maxage egen numlead = sum(lead), by(hid) gen rlead = runiform() egen maxrlead = max(rlead) if lead, by(hid) replace lead = 0 if numlead>1 & rlead<maxrlead *create variables for education* gen leadeduc_a=educ*lead egen leadeduc=max(leadeduc_a), by(hid) recode leadeduc (3=1) (nonmiss=0), gen(highed) recode leadeduc (1=1)(nonmiss=0), gen(lowed) gen agelead_a=age*lead egen agelead=max(agelead_a), by(hid) gen ageleadsq=agelead^2 *create family structure variables* gen married=. replace married=0 if marital>=200 & marital!=. replace married=1 if marital<200 partner==110 gen marriedhh_a=married*head egen marriedhh=max(marriedhh_a), by(hid) recode marriedhh (1=0)(0=1)(.=.), gen(single) *** singmom (based on nchildren & ageyoch) recode nchildren 2/17=1, gen(nchild) replace nchild=0 if ageyoch>17 & ageyoch!=. gen sing_mom_a=head*female

4 gen sing_mom_b=sing_mom_a*single gen sing_mom_c=sing_mom_b*nchild replace sing_mom_c=0 if age>54 egen singmom=max(sing_mom_c), by(hid) replace singmom=1 if singmom>1 & singmom!=. gen sing_dad_a=head*male gen sing_dad_b=sing_dad_a*single gen sing_dad_c=sing_dad_b*nhhmem17 egen singdad =max(sing_dad_c), by(hid) replace singdad=1 if singdad>1 gen fhnk_a=0 replace fhnk_a=1 if sing_mom_b ==1 & nhhmem17==0 egen fhnk=max(fhnk_a), by(hid) gen mhnk_a=0 replace mhnk_a=1 if sing_dad_b ==1 & nhhmem17==0 egen mhnk=max(mhnk_a), by(hid) save $mydata/`x', replace } ************************ *** build the Swiss file ************************ use $ch04h, clear drop if dhi==. drop if dhi==0 drop if hwgt==. gen ey=dhi/(sqrt(nhhmem)) qui sum ey gen botlin=0.01*_result(3) replace ey=botlin if ey<botlin quietly sum ey, de gen toplin=10*_result(10) sort hid keep iso2 did year hid hwgt nhhmem hhtype nhhmem65 nhhmem17 ey hi hil hile hic hicid hit dhi hpartner hitp hits hitsi hitsu hitsa nearn save $mydata/ch04h, replace *Person File use $ch04p, clear gen head=. // take head as the lead replace head=1 if relation==1000 replace head=0 if relation>1000 & relation!=. recode sex (1=0)(2=1)(.=.), gen(female)

5 recode sex (1=1)(2=0)(.=.), gen(male) sort hid keep hid pid did year hhmem relation partner children nchildren ageyoch age sex immigr yrsresid ethnic_c educ emp hours secjob inda1 indb1 indc1 sector1 occa1 occb1 pmi pmxit pmxiti ppension pi pil pmil pi pile pmile pils pmils pit pmit pits pmits pitsis pitsisun pmitsisun ptime gross1 educ_c marital head male female save $mydata/ch04p, replace merge m:1 hid using $mydata/brady/ch04h, keepusing (iso2 did year hid hwgt nhhmem hhtype nhhmem65 nhhmem17 ey hi hil hile hic hicid hit dhi hpartner hitp hits hitsi hitsu hitsa nearn) keep(match) nogen egen numemp=sum(emp), by(hid) replace nearn=numemp gen multearn=. replace multearn=0 if nearn==0 nearn==1 replace multearn=1 if nearn>1 & nearn!=. gen unemphh=. replace unemphh=0 if nearn>0 & nearn!=. replace unemphh=1 if nearn==0 *create variables for education* gen leadeduc_a=educ*head egen leadeduc=max(leadeduc_a), by(hid) recode leadeduc (3=1) (nonmiss=0), gen(highed) recode leadeduc (1=1)(nonmiss=0), gen(lowed) gen agelead_a=age*head egen agelead=max(agelead_a), by(hid) gen ageleadsq=agelead^2 *create family structure variables* gen married=. replace married=0 if marital>=200 & marital!=. replace married=1 if marital<200 partner==110 gen marriedhh_a=married*head egen marriedhh=max(marriedhh_a), by(hid) recode marriedhh (1=0)(0=1)(.=.), gen(single) *** singmom (based on nchildren & ageyoch) recode nchildren 2/17=1, gen(nchild) replace nchild=0 if ageyoch>17 & ageyoch!=. gen sing_mom_a=head*female gen sing_mom_b=sing_mom_a*single gen sing_mom_c=sing_mom_b*nchild replace sing_mom_c=0 if age>54 egen singmom=max(sing_mom_c), by(hid) replace singmom=1 if singmom>1 & singmom!=.

6 gen sing_dad_a=head*male gen sing_dad_b=sing_dad_a*single gen sing_dad_c=sing_dad_b*nhhmem17 egen singdad =max(sing_dad_c), by(hid) replace singdad=1 if singdad>1 gen fhnk_a=0 if female!=. & head!=. & nhhmem17!=. replace fhnk_a=1 if sing_mom_b ==1 & nhhmem17==0 egen fhnk=max(fhnk_a), by(hid) gen mhnk_a=0 if female!=. & head!=. & nhhmem17!=. replace mhnk_a=1 if sing_dad_b ==1 & nhhmem17==0 egen mhnk=max(mhnk_a), by(hid) save $mydata/ch04, replace **************** **** HUNGARY 05 **************** use $hu05h, clear drop if dhi==. drop if dhi==0 drop if hwgt==. gen ey=dhi/(sqrt(nhhmem)) qui sum ey gen botlin=0.01*_result(3) replace ey=botlin if ey<botlin quietly sum ey, de gen toplin=10*_result(10) sort hid keep iso2 did year hid hwgt nhhmem hhtype nhhmem65 nhhmem17 ey hi hil hile hic hicid hit dhi hpartner hitp hits hitsi hitsu hitsa nearn save $mydata/hu05h, replace *Person File use $hu05p, clear gen head=. // take head as the lead replace head=1 if relation==1000 replace head=0 if relation>1000 & relation!=. recode sex (1=0)(2=1)(.=.), gen(female) recode sex (1=1)(2=0)(.=.), gen(male) sort hid keep hid pid did year hhmem relation partner children nchildren ageyoch age sex immigr yrsresid ethnic_c educ emp hours secjob inda1 indb1 indc1 sector1 occa1 occb1 pmi pmxit pmxiti ppension pi pil pmil pi pile pmile pils pmils pit pmit pits pmits pitsis pitsisun pmitsisun ptime gross1 educ_c marital head male female

7 save $mydata/hu05p, replace merge m:1 hid using $mydata/brady/hu05h, keepusing (iso2 did year hid hwgt nhhmem hhtype nhhmem65 nhhmem17 ey hi hil hile hic hicid hit dhi hpartner hitp hits hitsi hitsu hitsa nearn) keep(match) nogen egen numemp=sum(emp), by(hid) replace nearn=numemp gen multearn=. replace multearn=0 if nearn==0 nearn==1 replace multearn=1 if nearn>1 & nearn!=. gen unemphh=. replace unemphh=0 if nearn>0 & nearn!=. replace unemphh=1 if nearn==0 *create variables for education* recode educ 9=2, gen(edu) tab edu educ,m gen leadeduc_a=edu*head egen leadeduc=max(leadeduc_a), by(hid) recode leadeduc (3=1) (nonmiss=0), gen(highed) recode leadeduc (1=1)(nonmiss=0), gen(lowed) gen agelead_a=age*head egen agelead=max(agelead_a), by(hid) gen ageleadsq=agelead^2 *create family structure variables* gen married=. replace married=0 if marital>=200 & marital!=. replace married=1 if marital<200 partner==110 gen marriedhh_a=married*head egen marriedhh=max(marriedhh_a), by(hid) recode marriedhh (1=0)(0=1)(.=.), gen(single) *** singmom (based on nchildren & ageyoch) recode nchildren 2/17=1, gen(nchild) replace nchild=0 if ageyoch>17 & ageyoch!=. gen sing_mom_a=head*female gen sing_mom_b=sing_mom_a*single gen sing_mom_c=sing_mom_b*nchild replace sing_mom_c=0 if age>54 egen singmom=max(sing_mom_c), by(hid) replace singmom=1 if singmom>1 & singmom!=. gen sing_dad_a=head*male

8 gen sing_dad_b=sing_dad_a*single gen sing_dad_c=sing_dad_b*nhhmem17 egen singdad =max(sing_dad_c), by(hid) replace singdad=1 if singdad>1 gen fhnk_a=0 if female!=. & head!=. & nhhmem17!=. replace fhnk_a=1 if sing_mom_b ==1 & nhhmem17==0 egen fhnk=max(fhnk_a), by(hid) gen mhnk_a=0 if female!=. & head!=. & nhhmem17!=. replace mhnk_a=1 if sing_dad_b ==1 & nhhmem17==0 egen mhnk=max(mhnk_a), by(hid) save $mydata/hu05, replace ************************** *** append country files ************************** global d "at04 be00 ca10 ch04 cz10 dk10 ee10 fi10 fr10 de10 gr10 hu05 ie10 il10 it10 is10 jp08 lu10 nl10 no10 pl10 sk10 si10 kr06 es10 se05 uk10 us13" use $mydata/brady/au10, clear foreach x of global d { append using "$mydata/brady/`x'" } save $mydata/prevpen_ajs_ca, replace ********* create some variables *** lead age groups gen leadu25=0 replace leadu25=1 if agelead<25 & agelead~=. gen lead2534=0 replace lead2534=1 if agelead>24 & agelead<35 gen leado54=0 replace leado54=1 if agelead>54 & agelead~=. **** alternative age definition: bottom third= young egen young3_a=cut(agelead), group(3) recode young3_a 1 2=0 0=1, gen(young3) *** sample year: pre or post > economic crisis gen post08=. replace post08=0 if year<2007 replace post08=1 if year>2007

9 **** poverty gen thresh=did recode thresh (140= ) (190=94738) (192= ) (208=22630) (209= ) (210=10077) /// (235=7578) (237=7107) (240=8326) (241=6525) (245= ) (247=20998) (251=3798.5) /// (252= ) (253=11160) (255=12025) (256= ) (259= ) (261=6451.5) /// (265= ) (267= ) (269=10716) (271=49670) (274=18356) (278=115334) /// (287= ) (289=109964) (293=632750) (295= ) (300= ) tab did thresh gen poor5=. replace poor5=0 if ey>=thresh & ey!=. replace poor5=1 if ey<thresh save $mydata/prevpen_ajs_ca, replace

10 BFH_LIS_2_figs1to3.do * Replication file for Brady, Finnigan, and Huebgen. "Rethinking the Risks of Poverty: A Framework for Analyzing Prevalences and Penalties." /* This do file contains the Stata code for obtaining the underlying estimates in Figures 1-3 of the article. The following code can be submitted to the LISSY interface of the Luxembourg Income Study ( by registered users. Please see the instructions (BFH_replication_instructions.txt) in this package for more information. The results from LISSY were copied into an excel file ("BFH_LISSY_results.xls"). The original version of the code in this file was written by Sabine Huebgen, Nov 10, Some edits by Ryan Finnigan, January */ *** Figure 1: Prevalences of the Four Risks of Poverty in 29 Rich Democracies (y-axis: percent of population). *** // loading data constructed in BFH_1_LIS_data.do use $mydata/brady/prevpen_ajs_ca, clear // estimating prevalences tabstat poor5 leadu25 lead2534 leado54 singmom fhnk mhnk nhhmem17 nhhmem65 lowed highed unemphh multearn [aw=hwgt] if agelead<65, by(did) stats (mean sd n) case /* results were copied into a sheet named "Fig1_3" in the excel file "BFH_LISSY_results.xlsx" Column names for the prevalences: did: LIS country codes id: country abbreviations wprevyoung: young headship (mean of "leadu25" above) wprevsingmom: single motherhood (mean of "singmom" above) wprevlowed: low education (mean of "lowed" above) wprevunemp: unemployed HH (mean of "unemphh" above) */ *** Figure 2: The Sum of Prevalences of Risks of Poverty in 29 Rich Democracies (x-axis: percent of population). *** // generating variable for the number of risks for each HH egen risksum=rowtotal(leadu25 singmom lowed unemphh), missing // generating separate dummy variables for having each number of risks (0 through 4, variables numbered 1 through 5) tab risksum, gen(prevsum) // looping over each dummy variable for number of risks, and estimating proportions with that number of risks by country forvalues i = 1/5 { tab did prevsum`i' [aw=hwgt] if agelead<65, row nof } // results were copied into a sheet named "Fig2" in the excel file "BFH_LISSY_results.xlsx" // columns rename "prevsum1" as "prevsum0", "prevsum2" as "prevsum1", and so on.

11 *** Figure 3: Penalties for the Four Risks of Poverty in 29 Rich Democracies (y-axis: increased probability of poverty). *** // looping through countries levelsof did, local(countries) foreach i of local countries { // estimating penalties with coefficients from linear probability models di "did = `i'" regress poor5 i.leadu25 i.lead2534 i.leado54 i.singmom i.fhnk i.mhnk nhhmem17 nhhmem65 i.lowed i.highed i.unemphh i.multearn [pw=hwgt] if agelead<65 & did==`i', cluster(hid) } /* results were copied into a sheet named "Fig1_3" in the excel file "BFH_LISSY_results.xlsx" Column names for the penalties: wpenyoung: young headship (coefficient for "leadu25" above) wpensingmom: single motherhood (coefficient for "singmom" above) wpenlowed: low education (coefficient for "lowed" above) wpenunemp: unemployed HH (coefficient for "unemphh" above) sigyoung: 0/1 for p<0.05 on "leadu25" coefficient sigsingmom: 0/1 for p<0.05 on "singmom" coefficient siglowed: 0/1 for p<0.05 on "lowed" coefficient sigunemp: 0/1 for p<0.05 on "unemphh" coefficient */

12 BFH_LIS_3_figs6_7.do * Replication file for Brady, Finnigan, and Huebgen. "Rethinking the Risks of Poverty: A Framework for Analyzing Prevalences and Penalties." /* This do file contains the Stata code for obtaining the underlying estimates in Figures 6 & 7 of the article. The following code can be submitted to the LISSY interface of the Luxembourg Income Study ( by registered users. Please see the instructions (BFH_replication_instructions.txt) in this package for more information. The results from LISSY were copied into an excel file ("BFH_LISSY_results.xlsx"). The original version of the code in this file was written by Sabine Huebgen, Nov 10, Some edits by Ryan Finnigan, January */ *** Figure 6: Counterfactual Simulation of U.S. Poverty with Cross-National Median Prevalences and Penalties *** // loading data constructed in BFH_1_LIS_data.do use $mydata/brady/prevpen_ajs_ca, clear // set of covariates global cov "leadu25 lead2534 leado54 singmom fhnk mhnk nhhmem17 nhhmem65 lowed highed unemphh multearn" // count variable for number of missing values per observation egen mi=rowmiss($cov) if agelead<65 // keeping complete cases keep if mi==0 // getting coefficient estimates for the US regress poor5 i.leadu25 i.lead2534 i.leado54 i.singmom i.fhnk i.mhnk nhhmem17 nhhmem65 i.lowed i.highed i.unemphh i.multearn [pw=hwgt] if agelead<65 & did==300, robust cluster(hid) /* linear regression Number of obs = 122,257 F(12, 41401) = Prob > F = R-squared = Root MSE = (Std. Err. adjusted for 41,402 clusters in hid) Robust poor5 Coef. Std. Err. t P> t [95% Conf. Interval] leadu lead leado singmom fhnk mhnk nhhmem nhhmem lowed

13 1.highed unemphh multearn _cons */ // predicted values US 2013 gen povhat=(leadu25* ) + (lead2534* ) /// + (leado54* ) + (singmom* ) + (fhnk* ) + (mhnk* ) /// + (nhhmem17* ) + (nhhmem65* ) + (lowed* ) + (highed* ) /// + (unemphh* ) + (multearn* ) + ( ) if agelead<65 & did==300 ** simulations with counterfactual prevalences ** /* Median prevalences are from BFH_local.do singmom: NET young: ICL lowed: ISR unemp: SVN */ // What if the US had median prevalences for all 4 groups? gen mprev=( * ) + (lead2534* ) /// + (leado54* ) + ( * ) + (fhnk* ) + (mhnk* ) /// + (nhhmem17* ) + (nhhmem65* ) + ( * ) + (highed* ) /// + (.06602* ) + (multearn* ) + ( ) if agelead<65 & did==300 ttest povhat=mprev if agelead<65 & did==300 // What if the US had median singmom prevalence keeping the rest as is? gen smprev=(leadu25* ) + (lead2534* ) /// + (leado54* ) + ( * ) + (fhnk* ) + (mhnk* ) /// + (nhhmem17* ) + (nhhmem65* ) + (lowed* ) + (highed* ) /// + (unemphh* ) + (multearn* ) + ( ) if agelead<65 & did==300 ttest povhat=smprev if agelead<65 & did==300 // What if the US had median younghead prevalencekeeping the rest as is? gen yprev=( * ) + (lead2534* ) /// + (leado54* ) + (singmom* ) + (fhnk* ) + (mhnk* ) /// + (nhhmem17* ) + (nhhmem65* ) + (lowed* ) + (highed* ) /// + (unemphh* ) + (multearn* ) + ( ) if agelead<65 & did==300 ttest povhat=yprev if agelead<65 & did==300 // What if the US had median lowed prevalence keeping the rest as is? gen lprev=(leadu25* ) + (lead2534* ) /// + (leado54* ) + (singmom* ) + (fhnk* ) + (mhnk* ) /// + (nhhmem17* ) + (nhhmem65* ) + ( * ) + (highed* ) /// + (unemphh* ) + (multearn* ) + ( ) if agelead<65 & did==300 ttest povhat=lprev if agelead<65 & did==300 // What if the US had median unemp prevalence keeping the rest as is?

14 gen uprev=(leadu25* ) + (lead2534* ) /// + (leado54* ) + (singmom* ) + (fhnk* ) + (mhnk* ) /// + (nhhmem17* ) + (nhhmem65* ) + (lowed* ) + (highed* ) /// + (.06602* ) + (multearn* ) + ( ) if agelead<65 & did==300 ttest povhat=uprev if agelead<65 & did==300 ** simulations with counterfactual penalties ** /* median penalties for risk groups from BFH_local_2_simulations.do singmom: young: lowed: unemp: */ // What if the US had median penalties for all 4 groups? gen mslope=(leadu25* ) + (lead2534* ) /// + (leado54* ) + (singmom* ) + (fhnk* ) + (mhnk* ) /// + (nhhmem17* ) + (nhhmem65* ) + (lowed* ) + (highed* ) /// + (unemphh* ) + (multearn* ) + ( ) if agelead<65 & did==300 ttest povhat=mslope if agelead<65 & did==300 // What if the US had median singmom penalty keeping the rest as is? gen smslope=(leadu25* ) + (lead2534* ) /// + (leado54* ) + (singmom* ) + (fhnk* ) + (mhnk* ) /// + (nhhmem17* ) + (nhhmem65* ) + (lowed* ) + (highed* ) /// + (unemphh* ) + (multearn* ) + ( ) if agelead<65 & did==300 ttest povhat=smslope if agelead<65 & did==300 // What if the US had median younghead penalty keeping the rest as is? gen yslope=(leadu25* ) + (lead2534* ) /// + (leado54* ) + (singmom* ) + (fhnk* ) + (mhnk* ) /// + (nhhmem17* ) + (nhhmem65* ) + (lowed* ) + (highed* ) /// + (unemphh* ) + (multearn* ) + ( ) if agelead<65 & did==300 ttest povhat=yslope if agelead<65 & did==300 // What if the US had median lowed penalty keeping the rest as is? gen lslope=(leadu25* ) + (lead2534* ) /// + (leado54* ) + (singmom* ) + (fhnk* ) + (mhnk* ) /// + (nhhmem17* ) + (nhhmem65* ) + (lowed* ) + (highed* ) /// + (unemphh* ) + (multearn* ) + ( ) if agelead<65 & did==300 ttest povhat=lslope if agelead<65 & did==300 // What if the US had median unemp penalty keeping the rest as is? gen uslope=(leadu25* ) + (lead2534* ) /// + (leado54* ) + (singmom* ) + (fhnk* ) + (mhnk* ) /// + (nhhmem17* ) + (nhhmem65* ) + (lowed* ) + (highed* ) /// + (unemphh* ) + (multearn* ) + ( ) if agelead<65 & did==300 ttest povhat=uslope if agelead<65 & did==300

15 *** Figure 7: Counterfactual Simulation of U.S. Poverty with Prevalences from 1970 and *** /* US 2013 weighted prevalences: singmom: young: lowed: unemp: */ // loading US data use if did==300 using $mydata/brady/prevpen_ajs_ca, clear // set of covariates global cov "leadu25 lead2534 leado54 singmom fhnk mhnk nhhmem17 nhhmem65 lowed highed unemphh multearn" // count variable for number of missing values per observation egen mi = rowmiss($cov) if agelead<65 // keeping complete cases keep if mi==0 // predicting poverty regress poor5 i.leadu25 i.lead2534 i.leado54 i.singmom i.fhnk i.mhnk nhhmem17 nhhmem65 i.lowed i.highed i.unemphh i.multearn [pw=hwgt] if agelead<65, robust cluster(hid) // predicted poverty in 2013 gen povhat=(leadu25* ) + (lead2534* ) /// + (leado54* ) + (singmom* ) + (fhnk* ) + (mhnk* ) /// + (nhhmem17* ) + (nhhmem65* ) + (lowed* ) + (highed* ) /// + (unemphh* ) + (multearn* ) + ( ) if agelead<65 /* US 1970 prevalences, estimated by Ryan Finnigan using Current Population Survey data from 1970, accessed through Integrated Public Use Microdata Series (IPUMS) at the University of Minnesota: singmom: young: lowed: unemp: */ // What if the US had the 1970s prevalences for all 4 risk groups? gen prev70=( * ) + (lead2534* ) /// + (leado54* ) + ( * ) + (fhnk* ) + (mhnk* ) /// + (nhhmem17* ) + (nhhmem65* ) + ( * ) + (highed* ) /// + ( * ) + (multearn* ) + ( ) if agelead<65 ttest povhat=prev70 if agelead<65 // What if the US had the 1970s singmom prevalence keeping the rest as is? gen smprev70=(leadu25* ) + (lead2534* ) /// + (leado54* ) + ( * ) + (fhnk* ) + (mhnk* ) /// + (nhhmem17* ) + (nhhmem65* ) + (lowed* ) + (highed* ) /// + (unemphh* ) + (multearn* ) + ( ) if agelead<65

16 ttest povhat=smprev70 if agelead<65 // What if the US had the 1970s younghead prevalence keeping the rest as is? gen yprev70=( * ) + (lead2534* ) /// + (leado54* ) + (singmom* ) + (fhnk* ) + (mhnk* ) /// + (nhhmem17* ) + (nhhmem65* ) + (lowed* ) + (highed* ) /// + (unemphh* ) + (multearn* ) + ( ) if agelead<65 ttest povhat=yprev70 if agelead<65 // What if the US had the 1970s lowed prevalence keeping the rest as is? gen lprev70=(leadu25* ) + (lead2534* ) /// + (leado54* ) + (singmom* ) + (fhnk* ) + (mhnk* ) /// + (nhhmem17* ) + (nhhmem65* ) + ( * ) + (highed* ) /// + (unemphh* ) + (multearn* ) + ( ) if agelead<65 ttest povhat=lprev70 if agelead<65 // What if the US had the 1970s unemp prevalence keeping the rest as is? gen uprev70=(leadu25* ) + (lead2534* ) /// + (leado54* ) + (singmom* ) + (fhnk* ) + (mhnk* ) /// + (nhhmem17* ) + (nhhmem65* ) + (lowed* ) + (highed* ) /// + ( * ) + (multearn* ) + ( ) if agelead<65 ttest povhat=uprev70 if agelead<65 /* US 1980 prevalences, estimated by Ryan Finnigan using Current Population Survey data from 1970, accessed through Integrated Public Use Microdata Series (IPUMS) at the University of Minnesota: singmom: young: lowed: unemp: */ // What if the US had the 1980s prevalences for all 4 risk groups? gen prev80=( * ) + (lead2534* ) /// + (leado54* ) + ( * ) + (fhnk* ) + (mhnk* ) /// + (nhhmem17* ) + (nhhmem65* ) + ( * ) + (highed* ) /// + ( * ) + (multearn* ) + ( ) if agelead<65 ttest povhat=prev80 if agelead<65 // What if the US had the 1980s singmom prevalence keeping the rest as is? gen smprev80=(leadu25* ) + (lead2534* ) /// + (leado54* ) + ( * ) + (fhnk* ) + (mhnk* ) /// + (nhhmem17* ) + (nhhmem65* ) + (lowed* ) + (highed* ) /// + (unemphh* ) + (multearn* ) + ( ) if agelead<65 ttest povhat=smprev80 if agelead<65 // What if the US had the 1980s younghead prevalencekeeping the rest as is? gen yprev80=( * ) + (lead2534* ) ///

17 + (leado54* ) + (singmom* ) + (fhnk* ) + (mhnk* ) /// + (nhhmem17* ) + (nhhmem65* ) + (lowed* ) + (highed* ) /// + (unemphh* ) + (multearn* ) + ( ) if agelead<65 ttest povhat=yprev80 if agelead<65 // What if the US had the 1980s lowed prevalence keeping the rest as is? gen lprev80=(leadu25* ) + (lead2534* ) /// + (leado54* ) + (singmom* ) + (fhnk* ) + (mhnk* ) /// + (nhhmem17* ) + (nhhmem65* ) + ( * ) + (highed* ) /// + (unemphh* ) + (multearn* ) + ( ) if agelead<65 ttest povhat=lprev80 if agelead<65 // What if the US had the 1980s unemp prevalence keeping the rest as is? gen uprev80=(leadu25* ) + (lead2534* ) /// + (leado54* ) + (singmom* ) + (fhnk* ) + (mhnk* ) /// + (nhhmem17* ) + (nhhmem65* ) + (lowed* ) + (highed* ) /// + ( * ) + (multearn* ) + ( ) if agelead<65 ttest povhat=uprev80 if agelead<65

18 BFH_LIS_4_tables2_3.do * Replication file for Brady, Finnigan, and Huebgen. "Rethinking the Risks of Poverty: A Framework for Analyzing Prevalences and Penalties." /* This do file contains the Stata code for Tables 2 & 3 of the article. The following code can be submitted to the LISSY interface of the Luxembourg Income Study ( by registered users. Please see the instructions (BFH_replication_instructions.txt) in this package for more information. The results from LISSY were copied into an excel file ("BFH_LISSY_results.xls"). The original version of the code in this file was written by Sabine Huebgen, Nov 10, Some edits by Ryan Finnigan, January */ *** Table 2: Multilevel Linear Probability Models of Individual Risks in 29 Rich Democracies *** // loading data constructed in BFH_1_LIS_data.do use $mydata/brady/prevpen_ajs_ca, clear // draw a random sample of all countries (4248= smallest country sample) keep if did<. & poor5<. & unemphh<. & multearn<. & lowed<. & highed<. & singmom<. & fhnk<. & mhnk<. & leadu25<. & lead2534<. & leado54<. & nhhmem17<. & nhhmem65<. & agelead<65 set seed sample 4248, count by(did) // set of covariates global cov "leadu25 lead2534 leado54 singmom fhnk mhnk nhhmem17 nhhmem65 lowed highed unemphh multearn" // inputting penalties from the single country analyses (see fig. 3) gen penleadu25=. replace penleadu25= if did==140 replace penleadu25= if did==190 replace penleadu25= if did==192 replace penleadu25= if did==208 replace penleadu25= if did==209 replace penleadu25= if did==210 replace penleadu25= if did==235 replace penleadu25= if did==237 replace penleadu25= if did==240 replace penleadu25= if did==241 replace penleadu25= if did==245 replace penleadu25= if did==247 replace penleadu25= if did==251 replace penleadu25= if did==252 replace penleadu25= if did==253 replace penleadu25= if did==255 replace penleadu25= if did==256 replace penleadu25= if did==259 replace penleadu25= if did==261 replace penleadu25= if did==265 replace penleadu25= if did==267 replace penleadu25= if did==269

19 replace penleadu25= if did==271 replace penleadu25= if did==274 replace penleadu25= if did==278 replace penleadu25= if did==287 replace penleadu25= if did==289 replace penleadu25= if did==295 replace penleadu25= if did==300 gen pensingmom=. replace pensingmom= if did==140 replace pensingmom= if did==190 replace pensingmom= if did==192 replace pensingmom= if did==208 replace pensingmom= if did==209 replace pensingmom= if did==210 replace pensingmom= if did==235 replace pensingmom= if did==237 replace pensingmom= if did==240 replace pensingmom= if did==241 replace pensingmom= if did==245 replace pensingmom= if did==247 replace pensingmom= if did==251 replace pensingmom= if did==252 replace pensingmom= if did==253 replace pensingmom= if did==255 replace pensingmom= if did==256 replace pensingmom= if did==259 replace pensingmom= if did==261 replace pensingmom= if did==265 replace pensingmom= if did==267 replace pensingmom= if did==269 replace pensingmom= if did==271 replace pensingmom= if did==274 replace pensingmom= if did==278 replace pensingmom= if did==287 replace pensingmom= if did==289 replace pensingmom= if did==295 replace pensingmom= if did==300 gen penlowed=. replace penlowed= if did==140 replace penlowed= if did==190 replace penlowed= if did==192 replace penlowed= if did==208 replace penlowed= if did==209 replace penlowed= if did==210 replace penlowed= if did==235 replace penlowed= if did==237 replace penlowed= if did==240 replace penlowed= if did==241 replace penlowed= if did==245 replace penlowed= if did==247 replace penlowed= if did==251 replace penlowed= if did==252 replace penlowed= if did==253 replace penlowed= if did==255

20 replace penlowed= if did==256 replace penlowed= if did==259 replace penlowed= if did==261 replace penlowed= if did==265 replace penlowed= if did==267 replace penlowed= if did==269 replace penlowed= if did==271 replace penlowed= if did==274 replace penlowed= if did==278 replace penlowed= if did==287 replace penlowed= if did==289 replace penlowed= if did==295 replace penlowed= if did==300 gen penunemp=. replace penunemp= if did==140 replace penunemp= if did==190 replace penunemp= if did==192 replace penunemp= if did==208 replace penunemp= if did==209 replace penunemp= if did==210 replace penunemp= if did==235 replace penunemp= if did==237 replace penunemp= if did==240 replace penunemp= if did==241 replace penunemp= if did==245 replace penunemp= if did==247 replace penunemp= if did==251 replace penunemp= if did==252 replace penunemp= if did==253 replace penunemp= if did==255 replace penunemp= if did==256 replace penunemp= if did==259 replace penunemp= if did==261 replace penunemp= if did==265 replace penunemp= if did==267 replace penunemp= if did==269 replace penunemp= if did==271 replace penunemp= if did==274 replace penunemp= if did==278 replace penunemp= if did==287 replace penunemp= if did==289 replace penunemp= if did==295 replace penunemp= if did==300 // 1st column: young head eststo m1: xtmixed leadu25 penleadu25 lowed highed unemphh multearn singmom fhnk mhnk nhhmem17 nhhmem65 if agelead<65 did:, robust // 2nd column: young head - eligible sample gen yeligible=0 replace yeligible=1 if highed==0 & age>17 & age<25 eststo m2: xtmixed leadu25 penleadu25 unemphh multearn lowed singmom fhnk mhnk nhhmem17 nhhmem65 if yeligible==1 did:, robust

21 // 3rd column: sing moms eststo m3: xtmixed singmom pensingmom highed lowed unemphh multearn leadu25 lead2534 leado54 nhhmem17 nhhmem65 if agelead<65 did:, robust // 4th column: sing moms - eligible sample gen smeligible=0 replace smeligible=1 if age<18 replace smeligible=1 if sex==2 & age<55 replace smeligible=0 if multearn==1 eststo m4: xtmixed singmom pensingmom unemphh lowed highed leadu25 lead2534 nhhmem17 nhhmem65 if smeligible==1 did:, robust // 5th column: lowed eststo m5: xtmixed lowed penlowed unemphh multearn singmom fhnk mhnk leadu25 lead2534 leado54 nhhmem17 nhhmem65 if agelead<65 did:, robust // 6th column: lowed - eligible sample gen leeligible=0 replace leeligible=1 if age>24 & age<55 & multearn==0 eststo m6: xtmixed lowed penlowed unemphh singmom fhnk mhnk leadu25 lead2534 leado54 nhhmem17 nhhmem65 if leeligible==1 did:, robust // 7th column: unemp hh eststo m7: xtmixed unemphh penunemp lowed highed singmom fhnk mhnk leadu25 lead2534 leado54 nhhmem17 nhhmem65 if agelead<65 did:, robust // 8th column: unemp hh - eligible sample gen uneligible=0 replace uneligible=1 if highed==0 & age>24 & age<55 eststo m8: xtmixed unemphh penunemp lowed singmom fhnk mhnk leadu25 lead2534 leado54 nhhmem17 nhhmem65 if uneligible==1 did:, robust // producing table 2 esttab m1 m2 m3 m4 m5 m6 m7 m8, obslast cons star(* 0.05 ** 0.01 *** 0.001) /// mtitles ("Young Head" "Young Head" "Single Mother HH" "Single Mother HH" "Low Education Head" "Low Education Head" "Unemployed HH" "Unemployed HH") /// keep(penleadu25 pensingmom penunemp penlowed unemphh lowed singmom leadu25 lead2534 leado54 highed fhnk mhnk nhhmem17 nhhmem65 multearn _cons) /// coeflabels(unemphh "Unemployed HH" leadu25 "Young Head" lowed "Low Educated Head" singmom "Single Mother HH" lead2534 "Head 25-34" leado54"head>54" /// fhnk "Female Head No Children" mhnk "Male Head No Children" nhhmem17 "# Children" nhhmem65 "#>65" highed "High Educated Head" multearn "Multiple Earner HH") *** Table 3: Multilevel Linear Probability Models of Poverty in 29 Rich Democracies (N=123,192) *** // inputting transfer share as z-score // transfer share data from: Brady, David and Amie Bostic "Paradoxes of Social Policy: Welfare Transfers, Relative Poverty, and Redistribution Preferences." American Sociological Review 80: gen zexten=did recode zexten /// (140= ) (190= ) (192= ) (208= ) (209= ) (210= ) /// (235= ) (237= ) (240= ) (241= ) (245= ) /// (247= ) (251= ) (252= ) (253= ) (255= ) /// (256=.27489) (259= ) (261= ) (265= ) (267= ) /// (269= ) (271= ) (274= ) (278= ) /// (287= ) (289= ) (295= ) (300= )

22 // inputting unemployment rates as a z-score // unemployment data from: Brady, David and Markus Jˮtti Ӆconomic Performance, Poverty, and Inequality in Rich Countries.ӠPp in The Oxford Handbook of the Social Science of Poverty, edited by D. Brady and L.M. Burton. New York: Oxford University Press. gen zunempr=did recode zunempr /// (140= ) (190= ) (192= ) (208= ) (209= ) /// (210= ) (235= ) (237= ) (240= ) (241= ) /// (245= ) (247= ) (251= ) (252= ) (253= ) /// (255= ) (256= ) (259= ) (261= ) (265= ) /// (267= ) (269= ) (271= ) (274= ) (278= ) /// (287= ) (289= ) (295= ) (300= ) // random intercept eststo m1: xtmixed poor5 zexten zunempr lowed highed unemphh multearn singmom fhnk mhnk leadu25 lead2534 leado54 nhhmem17 nhhmem65 if agelead<65 did:, robust // random slope for unemployed hh eststo m2: xtmixed poor5 c.zexten##unemphh zunempr lowed highed multearn singmom fhnk mhnk leadu25 lead2534 leado54 nhhmem17 nhhmem65 if agelead<65 did: unemphh, robust // random slope for low educated hh eststo m3: xtmixed poor5 c.zexten##lowed zunempr highed unemphh multearn singmom fhnk mhnk leadu25 lead2534 leado54 nhhmem17 nhhmem65 if agelead<65 did: lowed, robust // random slope for single motherhood eststo m4: xtmixed poor5 c.zexten##singmom zunempr lowed highed unemphh multearn fhnk mhnk leadu25 lead2534 leado54 nhhmem17 nhhmem65 if agelead<65 did: singmom, robust // random slope for young headship eststo m5: xtmixed poor5 c.zexten##leadu25 zunempr lowed highed unemphh multearn singmom fhnk mhnk lead2534 leado54 nhhmem17 nhhmem65 if agelead<65 did: leadu25, robust // producing table 3 esttab m1 m2 m3 m4 m5, obslast cons star(* 0.05 ** 0.01 *** 0.001) /// mtitles ("Random Intercept" "Random Slope: Unemployed" "Random Slope: Low-Educated" "Random Slope: Single Mother" "Random Slope: Young Head") /// coeflabels(unemphh "Unemployed HH" leadu25 "Young Head" lowed "Low Educated Head" singmom "Single Mother HH" lead2534 "Head 25-34" leado54"head>54" /// fhnk "Female Head No Children" mhnk "Male Head No Children" nhhmem17 "# Children" nhhmem65 "#>65" highed "High Educated Head" multearn "Multiple Earner HH" /// zexten "Transfer Share" zunempr "Unemployment Rate")

23 BFH_LIS_5_appendix.do * Replication file for Brady, Finnigan, and Huebgen. "Rethinking the Risks of Poverty: A Framework for Analyzing Prevalences and Penalties." /* This do file contains the Stata code for the article's appendix. Most of the following should be submitted to the LISSY interface of the Luxembourg Income Study ( by registered users. Please see the instructions (BFH_replication_instructions.txt) in this package for more information. Portions of the results from LISSY were copied into an excel file ("BFH_LISSY_results.xlsx"). The original version of the code in this file was written by Sabine Huebgen, Nov 10, This code has not been further edited to separate code submitted to LIS vs run locally, similar to the main replication files. All code should be submitted to LISSY unless noted in the comments (i.e., step 5 in generating Appendix III). */ *** Appendix I: Comparison of Penalties: Coefficients from Linear Probability Models and Average Marginal Effects from Logit Models for Four Risks in 29 High Income Democracies ** // loading data constructed in BFH_1_LIS_data.do use $mydata/brady/prevpen_ajs_ca, clear // new variable for countries sort did gen cntry=did replace cntry=1 if did==140 // BE replace cntry=2 if did==190 // SE replace cntry=3 if did==192 // KO replace cntry=4 if did==208 // CH replace cntry=5 if did==209 // HUN replace cntry=6 if did==210 // AT replace cntry=7 if did==235 // IT replace cntry=8 if did==237 // ES replace cntry=9 if did==240 // UK replace cntry=10 if did==241 // GR replace cntry=11 if did==245 // JP replace cntry=12 if did==247 // AU replace cntry=13 if did==251 // SK replace cntry=14 if did==252 // DE replace cntry=15 if did==253 // IE replace cntry=16 if did==255 // FI replace cntry=17 if did==256 // LU replace cntry=18 if did==259 // IL replace cntry=19 if did==261 // SI replace cntry=20 if did==265 // NE replace cntry=21 if did==267 // NO replace cntry=22 if did==269 // PL replace cntry=23 if did==271 // EE replace cntry=24 if did==274 // CA replace cntry=25 if did==278 // DK replace cntry=26 if did==287 // IS replace cntry=27 if did==289 // CZE replace cntry=28 if did==295 // FRA replace cntry=29 if did==300 // US

24 // calculate the penalties as AMEs for all countries forvalues i=1/29 { logit poor5 i.leadu25 i.lead2534 i.leado54 i.singmom i.fhnk i.mhnk nhhmem17 nhhmem65 i.lowed i.highed i.unemphh i.multearn [pw=hwgt] if agelead<65 & cntry==`i', cluster(hid) margins, dydx (*) } // values copied into BFH_LISSY_results.xlsx ***************************************** ********* App II: US models ********* ***************************************** use $mydata/brady/prevpen_ajs_ca, clear keep if did==300 *** first column: US model for calculating penalties (LPM) eststo m1: regress poor5 i.leadu25 i.lead2534 i.leado54 i.singmom i.fhnk i.mhnk nhhmem17 nhhmem65 i.lowed i.highed i.unemphh i.multearn [pw=hwgt] if agelead<65, cluster(hid) *** second column: US model for calculating penalties (AME) logit poor5 i.leadu25 i.lead2534 i.leado54 i.singmom i.fhnk i.mhnk nhhmem17 nhhmem65 i.lowed i.highed i.unemphh i.multearn [pw=hwgt] if agelead<65, cluster(hid) margins, dydx(*) *** third column: US model with race (LPM) recode ethnic_c 3= =3 7= =5 if did==300, gen (race) tab ethnic_c race tab race, gen (race) rename race1 white rename race2 black rename race3 hispa rename race4 asian rename race5 other eststo m2: regress poor5 i.leadu25 i.lead2534 i.leado54 i.singmom i.fhnk i.mhnk nhhmem17 nhhmem65 i.lowed i.highed i.unemphh i.multearn i.race [pw=hwgt] if agelead<65, cluster(hid) *** fourth column: US model with race and race-risk-interaction (LPM) eststo m3: regress poor5 i.leadu25 i.lead2534 i.leado54 i.singmom i.fhnk i.mhnk nhhmem17 nhhmem65 i.lowed i.highed i.unemphh i.multearn /// i.race i.black#unemphh i.hispa#i.unemphh i.black#i.leadu25 i.hispa#i.leadu25 i.black#i.singmom i.hispa#i.singmom i.black#i.lowed i.hispa#i.lowed [pw=hwgt] if agelead<65, cluster(hid) esttab m1 m2 m3, obslast cons star(* 0.05 ** 0.01 *** 0.001) /// mtitles ("Linear Probability: Coefficients and (T-Scores)" "Linear Probability, With Race: Coefficients and (T- Scores)" "Linear Probability, With Race and Race*Risk: Coefficients and (T-Scores)") /// keep(1.singmom 1.leadu25 1.lowed 1.unemphh 1.highed 1.lead leado54 1.multearn 1.fhnk 1.mhnk nhhmem17 nhhmem65 1.black 1.hispa 1.asian 1.other ///

25 1.black#1.unemphh 1.hispa#1.unemphh 1.black#1.lowed 1.hispa#1.lowed 1.black#1.singmom 1.hispa#1.singmom 1.black#1.leadu25 1.hispa#1.leadu25 /// coeflabels(unemphh "Unemployed HH" leadu25 "Young Head" lowed "Low Educated Head" singmom "Single Mother HH" lead2534 "Head 25-34" leado54"head>54" /// fhnk "Female Head No Children" mhnk "Male Head No Children" nhhmem17 "# Children" nhhmem65 "#>65" highed "High Educated Head" multearn "Multiple Earner HH") *************************************************** *** App III : Figure 6 with population size weights *************************************************** use $mydata/brady/prevpen_ajs_ca, clear *** step 1: calculate population size weighted penatlies without US *** generate population size (OECD --> gen popsize=. replace popsize= 103 if did==140 replace popsize= 90 if did==190 replace popsize= 484 if did==192 replace popsize= 74 if did==208 replace popsize= 101 if did==209 replace popsize= 82 if did==210 replace popsize= 605 if did==235 replace popsize= 466 if did==237 replace popsize= 619 if did==240 replace popsize= 112 if did==241 replace popsize= 1277 if did==245 replace popsize= 220 if did==247 replace popsize= 54 if did==251 replace popsize= 818 if did==252 replace popsize= 46 if did==253 replace popsize= 54 if did==255 replace popsize= 5 if did==256 replace popsize= 76 if did==259 replace popsize= 20 if did==261 replace popsize= 166 if did==265 replace popsize= 49 if did==267 replace popsize= 385 if did==269 replace popsize= 13 if did==271 replace popsize= 341 if did==274 replace popsize= 55 if did==278 replace popsize= 3 if did==287 replace popsize= 105 if did==289 replace popsize= 629 if did==295 replace popsize= 3165 if did==300 gen weight=popsize*hwgt global cov "leadu25 lead2534 leado54 singmom fhnk mhnk nhhmem17 nhhmem65 lowed highed unemphh multearn" egen mi=rowmiss($cov) if agelead<65

26 regress poor5 i.leadu25 i.lead2534 i.leado54 i.singmom i.fhnk i.mhnk nhhmem17 nhhmem65 i.lowed i.highed i.unemphh i.multearn [pw=weight] if agelead<65 & did!=300 &mi==0, robust cluster(hid) /* Number of obs = 1,140,895 F(12, ) = Prob > F = R-squared = Root MSE = (Std. Err. adjusted for 191,627 clusters in hid) Robust poor5 Coef. Std. Err. t P> t [95% Conf. Interval] leadu lead leado singmom fhnk mhnk nhhmem nhhmem lowed highed unemphh multearn _cons mean median penalties: - singmom: young: lowed: unemp:.2797 */ ** Step 2: Use these penalties for the counterfactual simulations * 1.1 What if the US had median penalties for all 4 groups? use $mydata/brady/prevpen_ajs_ca, clear keep if did==300 global cov "leadu25 lead2534 leado54 singmom fhnk mhnk nhhmem17 nhhmem65 lowed highed unemphh multearn" egen mi=rowmiss($cov) if agelead<65 regress poor5 i.leadu25 i.lead2534 i.leado54 i.singmom i.fhnk i.mhnk nhhmem17 nhhmem65 i.lowed i.highed i.unemphh i.multearn [pw=hwgt] if agelead<65 &mi==0, robust cluster(hid) *predicted values US 2013*

27 gen povhat=(leadu25* ) + (lead2534* ) /// + (leado54* ) + (singmom* ) + (fhnk* ) + (mhnk* ) /// + (nhhmem17* ) + (nhhmem65* ) + (lowed* ) + (highed* ) /// + (unemphh* ) + (multearn* ) + ( ) if agelead<65 * median slope for all 4 risk groups gen mslope=(leadu25* ) + (lead2534* ) /// + (leado54* ) + (singmom* ) + (fhnk* ) + (mhnk* ) /// + (nhhmem17* ) + (nhhmem65* ) + (lowed* ) + (highed* ) /// + (unemphh*.2797) + (multearn* ) + ( ) if agelead<65 * What if the US had median singmom penalty keeping the rest as is? gen smslope=(leadu25* ) + (lead2534* ) /// + (leado54* ) + (singmom* ) + (fhnk* ) + (mhnk* ) /// + (nhhmem17* ) + (nhhmem65* ) + (lowed* ) + (highed* ) /// + (unemphh* ) + (multearn* ) + ( ) if agelead<65 * What if the US had median younghead penalty keeping the rest as is? gen yslope=(leadu25* ) + (lead2534* ) /// + (leado54* ) + (singmom* ) + (fhnk* ) + (mhnk* ) /// + (nhhmem17* ) + (nhhmem65* ) + (lowed* ) + (highed* ) /// + (unemphh* ) + (multearn* ) + ( ) if agelead<65 * What if the US had median lowed penalty keeping the rest as is? gen lslope=(leadu25* ) + (lead2534* ) /// + (leado54* ) + (singmom* ) + (fhnk* ) + (mhnk* ) /// + (nhhmem17* ) + (nhhmem65* ) + (lowed* ) + (highed* ) /// + (unemphh* ) + (multearn* ) + ( ) if agelead<65 * What if the US had median unemp penalty keeping the rest as is? gen uslope=(leadu25* ) + (lead2534* ) /// + (leado54* ) + (singmom* ) + (fhnk* ) + (mhnk* ) /// + (nhhmem17* ) + (nhhmem65* ) + (lowed* ) + (highed* ) /// + (unemphh*.2797) + (multearn* ) + ( ) if agelead<65 ************************************************************ ** step 3: calculate population size weighted prevalences ************************************************************ use "$data/prevpen_lis_29.dta", clear tabstat leadu25 singmom lowed unemphh [pw= weight] if did!=300, stats(mean median) /* stats wpprev~g wpprev~m wpprev~d wpprev~p mean p mean median prevalences: