18
C I S May 2016 I n September 2015, the Center for Immigration Studies published a landmark study of immigration and wel- fare use, showing that 51 percent of immigrant-headed households used at least one federal welfare program — cash, food, housing, or medical care — compared to 30 percent of native households. Following similar methodology, this new study examines the dollar cost of that welfare use. e average household headed by an immigrant (legal or illegal) costs taxpayers $6,234 in federal welfare benefits, which is 41 percent higher than the $4,431 received by the average native household. e average immigrant household consumes 33 percent more cash welfare, 57 percent more food assis- tance, and 44 percent more Medicaid dollars than the average native household. Housing costs are about the same for both groups. At $8,251, households headed by immigrants from Central America and Mexico have the highest welfare costs of any sending region — 86 percent higher than the costs of native households. Illegal immigrant households cost an average of $5,692 (driven largely by the presence of U.S.-born chil- dren), while legal immigrant households cost $6,378. e greater consumption of welfare dollars by immigrants can be explained in large part by their lower level of education and larger number of children compared to natives. Over 24 percent of immigrant households are headed by a high school dropout, compared to just 8 percent of native households. In addition, 13 percent of immigrant households have three or more children, vs. just 6 percent of native households. Introduction In September 2015, the Center for Immigration Studies published a landmark study of immigration and welfare use, showing that 51 percent of immigrant-headed households (legal and illegal) use at least one federal welfare program, compared to 30 percent of native households. 1 “Welfare” refers to means-tested anti-poverty programs. ese include direct cash assistance in the form of Supplemental Security Income (SSI) and Temporary Assistance for Needy Families (TANF); food aid such as free school lunch, the Women, Infants, and Children (WIC) nutri- tion program, and food stamps; Medicaid; and housing assistance in the form of rent subsidies and public hous- ing. Not included are social insurance programs for which participants must generally pay into the system before drawing benefits, such as Social Security and Medicare. e earlier CIS study was notable for showing much higher welfare participation rates than previously reported. e reason is that earlier studies measured welfare participation with the Annual Social and Economic Supple- ment (ASEC) of the Current Population Survey. e ASEC is a simple cross-sectional dataset widely used in labor market research. However, the ASEC substantially undercounts welfare participation, in part because it asks respondents to recall their welfare use over a period between three and 15 months before the interview takes place. To address the undercount problem, CIS used a more complex dataset called the Survey of Income The Cost of Welfare Use By Immigrant and Native Households By Jason Richwine Jason Richwine, PhD, is an independent public policy analyst based in Washington, D.C., and a contributing writer at National Review. 1629 K Street, NW, Suite 600 • Washington, DC 20006 • (202) 466-8185 • [email protected] • www.cis.org

The Cost of Welfare Use By Immigrant and Native Householdstakes place. To address the undercount problem, CIS used a more complex dataset called the Survey of Income The Cost of Welfare

  • Upload
    others

  • View
    0

  • Download
    0

Embed Size (px)

Citation preview

Page 1: The Cost of Welfare Use By Immigrant and Native Householdstakes place. To address the undercount problem, CIS used a more complex dataset called the Survey of Income The Cost of Welfare

11629 K Street, NW, Suite 600, Washington, DC 20006 • Phone 202.466.8185 • Fax 202.466.8076 • www.cis.org

C I S

CIS Letterhead_Layout 1 7/26/12 4:34 PM Page 1

May 2016

In September 2015, the Center for Immigration Studies published a landmark study of immigration and wel-fare use, showing that 51 percent of immigrant-headed households used at least one federal welfare program — cash, food, housing, or medical care — compared to 30 percent of native households. Following similar

methodology, this new study examines the dollar cost of that welfare use.

• Theaveragehouseholdheadedbyanimmigrant(legalorillegal)coststaxpayers$6,234infederalwelfarebenefits,whichis41percenthigherthanthe$4,431receivedbytheaveragenativehousehold.

• Theaverageimmigranthouseholdconsumes33percentmorecashwelfare,57percentmorefoodassis-tance,and44percentmoreMedicaiddollarsthantheaveragenativehousehold.Housingcostsareaboutthe same for both groups.

• At$8,251,householdsheadedbyimmigrantsfromCentralAmericaandMexicohavethehighestwelfarecostsofanysendingregion—86percenthigherthanthecostsofnativehouseholds.

• Illegalimmigranthouseholdscostanaverageof$5,692(drivenlargelybythepresenceofU.S.-bornchil-dren),whilelegalimmigranthouseholdscost$6,378.

• Thegreaterconsumptionofwelfaredollarsbyimmigrantscanbeexplainedinlargepartbytheirlowerlevelofeducationand largernumberofchildrencomparedtonatives.Over24percentof immigranthouseholdsareheadedbyahighschooldropout,comparedtojust8percentofnativehouseholds.Inaddition,13percentofimmigranthouseholdshavethreeormorechildren,vs.just6percentofnativehouseholds.

IntroductionIn September 2015, the Center for Immigration Studies published a landmark study of immigration and welfare use,showingthat51percentofimmigrant-headedhouseholds(legalandillegal)useatleastonefederalwelfareprogram, compared to 30 percent of native households.1 “Welfare” refers to means-tested anti-poverty programs. TheseincludedirectcashassistanceintheformofSupplementalSecurityIncome(SSI)andTemporaryAssistanceforNeedyFamilies(TANF);foodaidsuchasfreeschoollunch,theWomen,Infants,andChildren(WIC)nutri-tionprogram,andfoodstamps;Medicaid;andhousingassistanceintheformofrentsubsidiesandpublichous-ing. Not included are social insurance programs for which participants must generally pay into the system before drawingbenefits,suchasSocialSecurityandMedicare.

TheearlierCISstudywasnotableforshowingmuchhigherwelfareparticipationratesthanpreviouslyreported.ThereasonisthatearlierstudiesmeasuredwelfareparticipationwiththeAnnualSocialandEconomicSupple-ment (ASEC)of theCurrentPopulationSurvey.TheASEC is a simple cross-sectionaldatasetwidelyused inlabormarket research.However, theASECsubstantiallyundercountswelfareparticipation, inpartbecause itasks respondents to recall their welfare use over a period between three and 15 months before the interview takesplace.Toaddresstheundercountproblem,CISusedamorecomplexdatasetcalledtheSurveyofIncome

The Cost of Welfare Use By Immigrant and Native Households

By Jason Richwine

Jason Richwine, PhD, is an independent public policy analyst based in Washington, D.C., and a contributing writer at National Review.

1629KStreet,NW,Suite600•Washington,DC20006•(202)466-8185•[email protected]•www.cis.org

Page 2: The Cost of Welfare Use By Immigrant and Native Householdstakes place. To address the undercount problem, CIS used a more complex dataset called the Survey of Income The Cost of Welfare

2

Center for Immigration Studies

and ProgramParticipation (SIPP).As the nameimplies, the Census Bureau specifically designed theSIPPtomeasureparticipationingovernmentprograms.Inaddition,theSIPPisa“longitudinal”dataset, meaning it follows the same respondents over time, asking them about their monthly pro-gram participation in three different interview “waves”throughouttheyear.Theresultisamuchmore complete picture of welfare participation comparedtowhattheASECprovides.

Table 1, adapted from thatCIS study, quantifiesthedifferences.WhiletheSIPPshowsthat51per-cent of immigrant-headed households and 30 per-cent of native-headed households used at least one welfare program in 2012, the comparable figures intheASECforimmigrantandnativehouseholdsarejust39percentand24percent,respectively.

NotethattheASECgenerallyundercountsimmi-grant welfare use more than it undercounts native use.Forexample,theMedicaidparticipationrateamongnativehouseholdsis1.27times(or27per-cent)higherintheSIPPcomparedtotheASEC,while Medicaid participation among immigranthouseholdsis1.39times(or39percent)higher.

Why Study the Cost of Welfare Use?Thecontri-butionofthisnewCISstudyistogobeyondparticipationratesintheSIPPbyestimatingthedollarcostsassociatedwithimmigrantandnativewelfareuse.Thepurposeistwo-fold.First,costestimatesareanaturalextensionoftheoriginalproject.WhileitisimportantforAmericanstounderstandtherateofwelfareuseamongimmigrants,expressingthatuseindollarterms offers a more tangible metric that is tied to current debates over fiscal policy. With the nation facing a long-term bud-getary deficit, this study helps illuminate immigration’s impact on the problem.

Thesecondpurposeismoretechnical.AselaboratedinthenextsectionandintheAppendix,astandardstrategyincoststudiesistotaketheundercountedcostsinsurveydataandadjustthemsothatthetotalequalstheofficialbudgetarynum-bers.Forexample,whentheNationalResearchCouncilconducteditscomprehensivefiscalanalysisofimmigrationin1997,thereport’sauthorsfirstcalculatedthepercentageofagivenwelfareprogram’scostattributedtoimmigrantsintheASEC,thenappliedthatpercentagetotheprogram’sofficialbudgetarycost.2TheassumptionwasthattheundercountofwelfareparticipationintheASECwasthesameforbothimmigrantsandnatives.3GivenTable1above,wenowknowthatassump-tiondoesnothold.UndercountisgreaterforimmigrantsintheASEC,meaningthatimmigrantsusemorewelfarerelativetonativesthanisreportedintheNationalResearchCouncilanalysis.EstimatingthedollarcostsofwelfareprogramsusingtheSIPPoffersthechanceforamorefine-tunedcomparison.

Methodology Outline.Thefederalbudgetshowsonlyhowmuchthegovernmentspendsoneachprogram,notthedemo-graphicsofrecipients.Therefore,theonlywaytoallocatewelfarecostsbetweenimmigrantsandnativesistostartwithasur-vey—inthiscase,theSIPP—thatincludesbothrespondents’welfareparticipationand their demographic characteristics.

TheSIPPcanbeusedtoestimatetheportionofwelfarecostslistedinthegovernment’sbudgetthatgoestoimmigrants.Theassumption here is notthatimmigrantscostexactlywhatisreportedintheSIPP,sincesurveysinevitablyundercountreceiptofgovernmentservices.ThekeyassumptionisonlythatthefractionofcostsattributedtoimmigrantsintheSIPPisthesameas the fraction of the real budgetary costs consumed by immigrants. For example, if immigrants account for 20 percent of SSIdollarsreportedintheSIPP,thisstudyassumesthatimmigrantsconsume20percentofactualSSIspendingreportedinthe budget.

Program

AnyWelfareCash

SSITANF

FoodSchool LunchWICSNAP

MedicaidHousing

PublicSubsidized

Table 1. Comparison of Welfare Participation Rates in the SIPP and the ASEC

SIPP

30.2%9.5%7.1%1.7%

21.8%12.4%4.2%

15.6%22.8%5.9%5.0%1.7%

SIPP

51.3%11.9%9.0%2.1%

40.3%30.0%10.9%20.8%41.6%6.0%5.0%1.6%

Native Households Immigrant Households

ASEC

24.0%5.3%4.2%1.3%

14.7%6.8%2.5%

10.7%17.9%4.3%3.0%1.3%

ASEC

38.5%6.3%4.5%2.0%

25.6%17.3%5.9%

13.5%29.9%

5.2%3.5%1.6%

Ratio

1.261.801.691.331.491.821.671.461.271.381.671.27

Ratio

1.331.881.991.051.581.731.841.541.391.151.430.98

Source: Camarota, “Welfare Use by Immigrant and Native Households”,TableA1.TheASECistheAnnualSocialandEconomicSupplementoftheCurrentPopulationSurvey.TheSIPPistheSurveyofIncomeandProgramParticipation.RatioistheSIPPparticipationratedividedbytheASECpar-ticipation rate. In this table, the Cash category includes several miscella-neous programs such as state general assistance and veterans’ compensa-tion.Intherestofthispaper,CashrefersexclusivelytoSSIandTANF.SeetheAppendixformoredetails.

Page 3: The Cost of Welfare Use By Immigrant and Native Householdstakes place. To address the undercount problem, CIS used a more complex dataset called the Survey of Income The Cost of Welfare

3

The above approach — calculating the costs at-tributable to immigrant- or native-headed house-holds in a survey, then adjusting those costs so that the total reflects official budgetarynumbers— isequivalenttotheNationalResearchCouncilmeth-od mentioned earlier, except that CIS uses the more accurateSIPPratherthantheASECtoestablishtheinitial allocation between immigrants and natives. PleaseseetheAppendixformoredetails.

Technicalspecificsaside,thefindingspresentedinthe next section are estimates.Allsurveys—eventhe best ones, such as the SIPP— are subject tomeasurement error, particularly when the surveys ask respondents for the amount of money they re-ceivefromvariousprograms.Adjustingthesurveycosts to reflect budgetary totals eliminates much of theuncertainty.However, theestimatespresentedin the next section should not be confused with ex-act budgetary figures.

FindingsThemain findings are presented in Table 2.Theaverage welfare cost in immigrant-headed house-holds is $6,234, compared to $4,431in native-headed households. Immi-grant households consume more cash, food,andMedicaiddollarsthannativehouseholds, while housing costs are roughly the same for both groups. Fig-ure1showsthatMedicaidisthelargestwelfare program, driving a large part of the overall difference between immi-grants and natives.

Sending Region. The cost of immi-grant welfare use varies by the immi-grants’ region of origin. Figure 2 shows that the highest-cost households are those headed by immigrants from Cen-tral America andMexico, consuminganaverageof$8,251inwelfarespend-ing.HouseholdsheadedbyimmigrantsfromEurope andAsia tend to be theleast costly. Unfortunately, individu-al countries are not identified in the SIPP,soafiner-grainedanalysis isnot possible.4

Legal Status.Although theSIPPdoesnot directly measure legal status, prob-ability models can be used to determine which immigrants are most likely to be in the country illegally.5 Table 3 indi-

Program

AnyWelfareCash

SSITANF

FoodSchool LunchWICSNAP

MedicaidHousing

Sample SizeWeightedn(millions)

Table 2. Average Cost of Household Welfare Use in 2012

Cost

4,43151744670

6896632

5902,831395

22,077104.6

Cost

6,23468658997

1,08317983

8214,072394

2,98016.16

Native Households

Immigrant Households Immigrant/

Native Cost Ratio

90 % C.I. (±)

14029278

2632

249926

90 % C.I. (±)

41393892789129

7927063

1.411.331.321.381.572.712.571.391.441.00

Source: Surveyof Income andProgramParticipation covering calendaryear2012,alongwithfederalbudgetdata.Householdsareclassifiedbythenativity of the household head. C.I. = confidence interval.

Figure 1. Immigrant households consumed more welfare than native households in 2012.

Source: SurveyofIncomeandProgramParticipationcoveringcalendaryear2012,alongwith federal budget data. Foodprogramsincludefreeorreducedschoollunch,WIC,andfoodstamps;cashin-cludesSSIandTANF;andhousingincludessubsidizedandpublichousing. Householdsareclassifiedbythenativityofthehouseholdhead.

7,000

Native Households

Immigrant Households

6,000

$4,431

$6,234

$517 $686 $689$1,083

$2,831

$4,072

$395 $394

5,000

4,000

3,000

2,000

1,000

Aver

age

Hou

seho

ld C

ost

0All

WelfareCash Food Medicaid Housing

Page 4: The Cost of Welfare Use By Immigrant and Native Householdstakes place. To address the undercount problem, CIS used a more complex dataset called the Survey of Income The Cost of Welfare

4

Center for Immigration Studies

cates that households headed by (likely) illegal immigrantshave an average welfare cost of $5,692. Illegal immigrants arebarred from directly accessing most (though not all) welfareprograms, but they can receive welfare through their U.S.-born children.6 Legal immi-grant households, which have greater eligibility for welfare, cost$6,378onaverage.

Workers. A popular miscon-ception about the Americanwelfare system is that it mainly benefits people who are not in the labor force. In fact, most means-tested anti-poverty programs are open to low-wage workers. For that reason, limiting the analysis to house-holds with at least one worker, asTable4does,onlymodestlyreduces the welfare cost esti-mates. The drop is especiallysmall for immigrant house-holds — from an overall cost of$6,234 inTable2 to$5,340in Table 4— because 84 per-cent of immigrant households already contain a worker (vs.73percentofnativehouseholds).Therefore,thehigher welfare spending on immigrant house-holds compared to native households is not due toa lackofworkamong immigrants.Thedif-ference is better explained by the demographic factors analyzed below.

Education. It is easy to understand why people with fewer skills are more likely to participate in welfare programs, since eligibility for those programsrequiresalowincome.Unsurprising-ly,Table5showsthatwelfarecostsdecreaseasthe education of the household head increases. Whereas immigrant households headed by a highschooldropoutcostanaverageof$10,329,immigrant households with college-educated headscost just$2,455.Thetabledemonstratesthat the difference between immigrant and na-tive household welfare costs becomes smaller after accounting for education.

Some differences remain. College-educated immigrant households consume substantially more welfare than comparably educated na-

Figure 2. Household welfare consumption in 2012 tended to be higher among immigrants from Latin America and lower among immigrants from Europe and Asia.

Source: SurveyofIncomeandProgramParticipationcoveringcalendaryear2012,alongwithfederal budget data.Householdsareclassifiedbythenativityofthehouseholdhead. Individual countries are not identified in the source data.

Natives

Central America and Mexico

$4,431

$8,251

$6,159

$5,705

$5,631

$5,260

$3,509

$2,565

South America

Caribbean

Africa

East Asia

Europe

South Asia

Program

AnyWelfareCash

SSITANF

FoodSchool LunchWICSNAP

MedicaidHousing

Sample SizeWeightedn(millions)

Table 3. Average Household Cost of Welfare Use in 2012, by Legal Status

Cost

4,43151744670

6896632

5902,831395

22,077104.6

Cost

6,37880571292

99914364

7924,131443

2,41212.77

Cost

5,692238122116

1,399314156929

3,8462095683.39

Native Households

Legal Imm.Households

Illegal Imm.Households

90 % C.I. (±)

14029278

2632

249926

90 % C.I. (±)

485113108

3195129

8531373

90 % C.I. (±)

669846554

1943325

16647684

Source:SurveyofIncomeandProgramParticipationcoveringcalendaryear2012,alongwithfederalbudgetdata.Householdsareclassifiedbythenativityof the household head. Legal status is determined by a probability model de-scribedinCamarota,“WelfareUsebyLegalandIllegalImmigrantHouseholds”.C.I. = confidence interval.

Page 5: The Cost of Welfare Use By Immigrant and Native Householdstakes place. To address the undercount problem, CIS used a more complex dataset called the Survey of Income The Cost of Welfare

5

tives.Attheotherendoftheskillspectrum, immigranthouse-holds headed by a high school dropout use less welfare than their nativecounterparts.Moreimportantthanthoseintra-educationdifferences, however, is the distribution of education across households.The“Percentage”rowsinTable5showthatover24percent of immigrant households are headed by a high-school dropout,vs.just8percentofnativehouseholds.Suchstarkedu-cational differences will inevitably lead to differences in welfare consumption.

Children. Another important cause of the difference betweenimmigrant and native households is the presence of minor chil-dren.Table6showsthatcostsaresimilarforagivennumberofchildren, and immigrant households have more children on aver-agethannativehouseholds.However,thepresenceofchildrenisnot the only reason for the relatively higher welfare cost of im-migranthouseholds.Table6 shows that immigranthouseholdswithout children consume significantly more welfare dollars than childless native households.

Race and Ethnicity. Regardless of nativity, households headed by blacks orHispanics consumemorewelfare dollars on aver-agethanhouseholdsheadedbywhitesorAsians.However,Table7 shows some interesting immigrant-native differences withinracialgroups.ImmigranthouseholdsheadedbyblacksandHis-panics cost less than their native counterparts, while white- and Asian-headed immigrant households cost more. Despite thelowercostamongimmigrantHispanicscomparedtonativeHis-panics, a much larger proportion of immigrant households are headedbyHispanics,whichcontributestothegreateroverallcostdifference between immigrants and natives.

Explaining Immigrant-Native Cost Differences with Regression Analysis. What explains the cost difference between im-migrantandnativehouseholds?Theprecedingsectionscontroloneatatimeforfactorssuchaseducationandnumberofchildren.Thissectionusesregressionanalysistosimultaneouslycontrolformultipleexplanatoryvariables,givingabettersense of which factors are most important.

Table8showshowthedifferencebetweenimmigrantandnativewelfarecostsvariesdependingonthecontrols.Thefirstrow gives the baseline estimate with no controls other than an indicator for immigrant status. In the no-control scenario, immigranthouseholdscost$1,803morethannativehouseholds,whichisconsistentwithTable2above.Thesecondrowshowsthattheimmigrant-nativedifferencebecomeslarger—upto$2,323—whenwecontrolforthepresenceofaworkerinthehousehold.Thedifferencethenbecomesgraduallysmallerascontrolsareaddedforeducationandnumberofchildren.Thefourthrowshowsthatimmigranthouseholdswiththesameworkerstatus,education,andnumberofchildrenasna-tivehouseholdscostjust$309more,whichisastatisticallyinsignificantdifference.Thefifthrowshowsthatimmigrantsusefewer welfare dollars when they are compared to natives of the same race as well as worker status, education, and number of children.

Althoughthisregressionanalysisisinterestingfromasociologicalperspective,itisimportantnottooverstateitsimportance.Once we control for demographic factors, we are now looking only at hypothetical costs that might exist if immigrants had the average characteristics of natives. It is easy to see how these hypothetical costs could be abused in policy debates. For ex-ample,immigrationrestrictionistsmightarguethatimmigrantsareuniquelypronetousingwelfarebecausetheycostmorethan natives with the same employment status. Supporters of expanded immigration might counter that immigrants’ greater attachment to the workforce is more important than their welfare use.

Similarly, immigration advocates often argue that although immigrants use more welfare than natives, low-skill immigrants use less welfare than low-skill natives.7(Seethe“lessthanhighschool”columninTable5.)Criticsrespondthatthisisthe

Program

AnyWelfareCash

SSITANF

FoodSchool LunchWICSNAP

MedicaidHousing

Sample SizeWeightedn(millions)

Table 4. Average Cost of Welfare Use in 2012 in Households with a Worker

Cost

3,20829125041

5967135

4902,135187

15,37276.7

Cost

5,34045838475

1,03519691

7483,638208

2,42713.64

Native Households

Immigrant Households

90 % C.I. (±)

12625248

2733

259318

90 % C.I. (±)

39183792685131075

27744

Source:SurveyofIncomeandProgramParticipationcov-ering calendar year 2012, along with federal budget data.Householdsareclassifiedbythenativityofthehouseholdhead.Worker status is determined by the January interview. C.I. = confidence interval.TableA5containsmoredetaileddata.

Page 6: The Cost of Welfare Use By Immigrant and Native Householdstakes place. To address the undercount problem, CIS used a more complex dataset called the Survey of Income The Cost of Welfare

6

Center for Immigration Studies

wrongstandardforjudgment.Justbecausepoorandunskilledimmigrantsmayconsumefewerwelfaredollarsthanequallypoor and unskilled natives, it does not necessarily follow that such immigrants are good candidates for residence and citi-zenship,sincetheUnitedStatescouldsimplypursueimmigrantswhodonothavelowlevelsofeducationinthefirstplace.

ThemostimportantrowinTable8isthefirstone,whichshowstheactualcostdifferencesbetweenimmigrantsandnatives.Theremainingrowshelpestablishthereasonsforthedifferences—and,insodoing,mayinformpolicydebatesregardingskill selection — but immigrants do not become more or less costly based on their hypothetical characteristics. It is the actual characteristics that matter.

Cost of Households Currently Using a Given Welfare Program. Up to thispoint, every table in this studyhas shownwelfare costs averaged across all households regardless of whether the households are participating in a welfare pro-gram. In other words, many $0values are included in theaverages.Thesetablesanswerthe question, “How muchdoes the average immigrant household cost in welfare ex-penditures compared to na-tive households?”

A related but different ques-tion is how much immigrant and native households cost once they are on the welfare program inquestion.Table9restricts the cost averages to households participating in the welfare program listed in each row. For example, among immigrant households receiv-ing SSI, the average benefit is $6,561, compared to $6,274for native households receiv-ing SSI. Immigrant and native household costs are generally similar in Table 9, with theimportant exceptionofMed-icaid.Onceenrolled inMed-icaid, immigrant households cost less on average than na-tive households, perhaps due to immigrant coverage being more tilted toward children rather than the elderly and disabled. In fact, because of legal restrictions, immigrant households are more likely to be “child-only” welfare recipients.8

Program

AnyWelfareCash

SSITANF

FoodSchool LunchWICSNAP

MedicaidHousing

Sample SizeWeightedn(millions)Percentage

Program

AnyWelfareCash

SSITANF

FoodSchool LunchWICSNAP

MedicaidHousing

Sample SizeWeightedn(millions)Percentage

Table 5. Average Cost of Household Welfare in 2012, by Education

Cost

11,6711,5731,395179

1,653141

501,4627,1691,2751,9888.157.8%

Cost

10,3291,067897170

1,946332159

1,4566,7765397563.93

24.3%

Cost

6,2027005821189898447

8573,941573

5,64525.8124.7%

Cost

7,829762577185

1,486239100

1,1475,0085736943.87

23.9%

Cost

4,49650944168

7367836

6222,885367

7,88538.4436.7%

Cost

5,08163760334

72912368

5383,4073096373.56

22.0%

Cost

1,099112104

71471911

11777961

6,55932.2030.8%

Cost

2,455349336

133144719

2481,5981948934.81

29.7%

Less Than High School

Less Than High School

Native Households

Immigrant Households

High School

High School

SomeCollege

SomeCollege

College DegreeOr More

College DegreeOr More

90 % C.I. (±)

80417416454

1301310

119532159

90 % C.I. (±)

87917917567

2152726

192658119

90 % C.I. (±)

4127065207176

6528658

90 % C.I. (±)

74515814273

1682524

149526133

90 % C.I. (±)

21742421643

54

4015238

90 % C.I. (±)

794270269

31153

2317

139542102

90 % C.I. (±)

11523224

2133

198914

90 % C.I. (±)

3979997128378

7926168

Source: Survey of Income andProgramParticipation covering calendar year 2012, alongwithfederal budget data. Householdsareclassifiedbythenativityofthehouseholdhead. C.I. = confidence interval. TableA5containsmoredetaileddata.

Page 7: The Cost of Welfare Use By Immigrant and Native Householdstakes place. To address the undercount problem, CIS used a more complex dataset called the Survey of Income The Cost of Welfare

7

The Broader Fiscal PictureThisstudyfocusesonthecostofmajorwelfareprogramsusedbyimmigrantandnativehouseholds.Bycontrast,acompletefiscal analysis would measure the cost of all government services and compare those costs with the taxes paid by each type of household. Some readers may wonder whether broadening the analysis would reveal that immigrant households make up for theirgreaterwelfarecostbypayinghighertaxes.Thisisnotthecase.AsthepreviousCISstudyofwelfareparticipationdem-onstrated, immigrant house-holdspayonlyabout89centsin federal income and payroll taxes for every dollar paid by native households.9

The aforementioned reportby the National Research Council, which did measure all government expenditures and taxes paid, found that im-migrant households cost tax-payersasmuchas$2,200peryear in the 1990s, dependingon their state of residence.10 More recently, the HeritageFoundation’s complete fiscal analysis (towhichtheauthorof this studycontributed)es-timated that the average legal immigrant household paid $4,344 less in taxes than itreceived in services in 2010, compared to a deficit of just $310 for the average nativehousehold.11 For the most up-to-date numbers, the National Research Council will release a new analysis later this year.

Thestudiesmentionedabovemeasure the direct fiscal effects of immigration by compar-ing the services households receive with the taxes they pay. But what about indirect effects? Immigration touches allaspectsofAmericanlife,soone could give almost endless examples of immigrants influ-encing society in ways that in-directly change how much the government taxes and spends. Attempting to quantify someof those indirect effects is not objectionable in itself, but itdoesopena“Pandora’sBox” of selectivity bias and exaggeration.

Program

AnyWelfareCash

SSITANF

FoodSchool LunchWICSNAP

MedicaidHousing

Sample SizeWeightedn(millions)Percentage

Program

AnyWelfareCash

SSITANF

FoodSchool LunchWICSNAP

MedicaidHousing

Sample SizeWeightedn(millions)Percentage

Table 6. Average Cost of Household Welfare Use in 2012, by Number of Children

Cost

2,62045243616

24533

2401,632291

15,83872.5969.4%

Cost

4,145820796

23374

56

3632,528423

1,6728.63

53.4%

Cost

6,115631472159

1,0829466

9223,849554

2,70613.94

13.3%

Cost

5,49039733958

91015188

6713,7324515002.99

18.5%

Cost

7,512584420165

1,46317887

1,1994,912

5522,23811.51

11.0%

Cost

7,467538362175

1,557303157

1,0985,0653084452.47

15.3%

Cost

15,461874554320

3,395513192

2,68910,271

9211,2956.576.3%

Cost

14,540723355368

3,717794307

2,6169,8072923632.07

12.8%

NoChildren

NoChildren

Native Households

Immigrant Households

OneChild

OneChild

TwoChildren

Two Children

Three or MoreChildren

Three or MoreChildren

90 % C.I. (±)

1233433

516

11

168123

90 % C.I. (±)

396139141

1551

23

5025574

90 % C.I. (±)

3738370357269

6727483

90 % C.I. (±)

678128120

331401726

128496147

90 % C.I. (±)

512947945

103111295

33696

90 % C.I. (±)

1,01418613984

2362730

209710125

90 % C.I. (±)

1,05413710082

2473124

218733161

90 % C.I. (±)

1,4822141581453524043

3321,150160

Source: Survey of Income andProgramParticipation covering calendar year 2012, alongwithfederal budget data. Householdsareclassifiedbythenativityofthehouseholdhead.Child status is determined by the January interview. Some children occasionally enter the house-hold after January, which explains why non-child households have non-zero costs for school lunch and WIC. C.I. = confidence interval. TableA6containsmoredetaileddata.

Page 8: The Cost of Welfare Use By Immigrant and Native Householdstakes place. To address the undercount problem, CIS used a more complex dataset called the Survey of Income The Cost of Welfare

8

Center for Immigration Studies

Forexample,considerthereactiontotheHeritageFoundation’sestimatethatillegalimmigrationandamnestywouldgen-erateadirectlifetimecostof$6.3trillion.Supportersofamnestyquicklysettledonarebuttalpoint:Althoughillegalim-migrantswhoreceiveamnestymaypayasagroup$6.3trillionlessintaxesthantheyreceiveinbenefitsovertheirlifetimes,their labor boosts economic productivity so much that natives probably still end up in the black.12Thatclaimis,firstofall,atremendousexaggeration.Mostofthegainsfromimmigrationgotoimmigrantsthemselves,nottonatives.13 In a paper forCISbackin2013,economistGeorgeBorjasestimatedthatillegalimmigrantsincreasedGDPby$395billionto$472bil-lion.Ofthatamount,however,onlyabout$9billionwenttonatives.14Afterextendingthat$9billionannuallyoveranadultlifetimeof50years,productivitygainswouldaddbackjust7percentofthe$6.3trillionfiscalcost.

Furthermore, an increase in productivity is just one of many indirect fiscal effects of immigration. What is the cost of addi-tional welfare spending on natives when they are displaced from jobs or see their wages lowered by immigrant competition?15 What are the moving and commuting costs incurred by natives who flee overcrowd-ing? What are the costs of less social trust and cooperation identified by Robert Putnamand others?16How about theincrease in English-languagelearners in public schools? One could go on and on with costs and benefits of immi-gration that indirectly impact the government’s fiscal situa-tion. But once advocates enter the world of indirect effects, they become decidedly selec-tive with the effects they wish to include.

ConclusionWhen researchers analyze welfare participation and costs, their dataset of choice has traditionally been the AnnualSocialandEconomicSupplement (ASEC) of theCurrent Population Survey.While the ASEC is certainlyuseful—CISusesitfrequent-ly — it substantially under-counts welfare participation. For that reason, CIS turned to the Survey of Income and ProgramParticipation(SIPP),a more complex dataset de-veloped by the Census Bureau specifically to analyze wel-fare use. CIS’s analysis of the SIPP has now generated twomajorstudies.Thefirststudy,published in September 2015, measured welfare participa-tion rates. It showed that 51

Program

AnyWelfareCash

SSITANF

FoodSchool LunchWICSNAP

MedicaidHousing

Sample SizeWeightedn(millions)Percentage

Program

AnyWelfareCash

SSITANF

FoodSchool LunchWICSNAP

MedicaidHousing

Sample SizeWeightedn(millions)Percentage

Table 7. Average Cost of Household Welfare Use in 2012, by Race

Cost

8,3751,010798212

1,39219777

1,1175,072901

1,2178.057.7%

Cost

7,863756575181

1,632317148

1,1675,163

3121,1577.08

43.8%

Cost

10,0731,2311,030

2011,702

15168

1,4835,8631,2762,71413.4012.8%

Cost

6,47046444124

1,03411057

8674,3785932721.549.5%

Cost

2,957333303

304333721

3752,007185

17,32278.9675.5%

Cost

4,56454950346

5717229

4702,9594858674.06

25.1%

Cost

3,183467464

317854

3121

2,373165176

1.031.0%

Cost

4,76480478816

5795026

5043,019362626

3.2019.8%

Hispanic

Hispanic

Native Households

Immigrant Households

Black

Black

White

White

Asian

Asian

90 % C.I. (±)

64615413660

1271713

114435161

90 % C.I. (±)

69217517256

1602319

13847272

90 % C.I. (±)

65514112342

1201110

111429118

90 % C.I. (±)

1,556183180

212452622

2271,120278

90 % C.I. (±)

1362424

524

22

2210017

90 % C.I. (±)

61513112224

104129

94394120

90 % C.I. (±)

1,237308308

36522

357

944137

90 % C.I. (±)

691146144

13149

1212

144446109

Source: Survey of Income andProgramParticipation covering calendar year 2012, alongwithfederal budget data. Householdsareclassifiedbythenativityofthehouseholdhead. C.I. = confidence interval. TheBlack,White,andAsiancolumnsexcludeHispanics.“Other”racenotshown. TableA6andTableA7containmoredetaileddata.

Page 9: The Cost of Welfare Use By Immigrant and Native Householdstakes place. To address the undercount problem, CIS used a more complex dataset called the Survey of Income The Cost of Welfare

9

percent of immigrant-headed households used some form of welfare, compared to 30 percent of native households.17ThissecondstudyextendstheSIPPanalysisbymovingfromratestocosts. It finds that immigrant-headed households consume anaverageof$6,234inwelfarespending,comparedto$4,431fornativehouseholds.Thehighest-costimmigranthouseholdstendtobethoseheadedbyapersonfromLatinAmerica,whilethelowest-costhouseholdsareheadedbypeoplefromEuropeandAsia.

Thisstudyimpliesthattwocompetingnarrativesaboutimmigrationareboth true. Immigrants do indeed have a strong at-tachmenttothelaborforce,asimmigrationadvocatesoftenpointout.Atthesametime,however,immigrantsconsumealargeamountofwelfarespending,justascriticsclaim.ThereasonthatbothnarrativesaretrueisthattheAmericanwelfaresystemhasbecomeincreasinglyfocusedonbuttressinglow-wageworkersratherthansupportingnon-workers.Putmoresimply, welfare and low-wage work go together. Just as natives with low levels of education and large numbers of children are apttoconsumewelfare,immigrantswiththosesamecharacteristicsarealsolikelytobeonwelfare.Astrongworkethicdoesnot change this reality.

Inordertoreducethecostofimmigrantwelfareuse,eitherthewelfaresystemortheimmigrationsystemmustchange.Theformer option is sometimes described as “building a wall around the welfare state” to prevent new immigrants from access-ing it. It is easier said than done. Loopholes and exceptions have weakened previous attempts to limit immigrant access to welfare.18Moreimportantly,CongresshasnopowertopreventtheU.S.-bornchildrenofimmigrantsfromusingthesamewelfare programs that the children of natives do. No matter how strong the “wall around the welfare state” is built, it cannot stopimmigrantparentsfromsigninguptheirU.S.-bornchildrenforMedicaid,SNAP,freeschoollunch,etc.,aslongasnativeparents can do the same.

Only a full-scale rollback of the welfare state for both immigrants and natives would prevent immigrant families from con-suming welfare dollars. Whatever one thinks of that proposal, it is not a policy change likely to occur in the near future.19 In fact, importing new clients of the welfare state likely makes it even harder to roll back.20AslongastheU.S.continuestoadmitlargenumbersoflow-skillimmigrants(legalorillegal),thenimmigrantwelfareconsumptionwillremainhigh.

Regression

1

2

3

4

5

Sample SizeWeightedn(millions)

Table 8. Immigrant Household Welfare Use Minus Native Household Welfare Use

Controls

None

Householdworker

HouseholdworkerNumber of children

HouseholdworkerNumber of childrenEducationofhead

HouseholdworkerNumber of childrenEducationofhead

Race of head

25,057120.8

CostDifference

1,803

2,323

1,196

309

-626

90 % C.I. (±)

446

435

401

427

472

Source:Source:SurveyofIncomeandProgramParticipationcover-ing calendar year 2012, along with federal budget data. Apositivevaluemeansthatimmigrantscostmorethancomparablenatives.Anegativevaluemeansimmigrantscostless. Householdsareclassifiedbythenativityofthehouseholdhead.C.I. = confidence interval.

Program

AnyWelfareCash

SSITANF

FoodSchool LunchWICSNAP

MedicaidHousing

Table 9. Average Cost of Households Currently Using Given Welfare Program

Cost

14,9546,2666,2744,0703,190

535755

3,88612,4196,672

Cost

12,2176,5456,5614,6372,686597751

4,0239,7936,604

Native Households

Immigrant Households

90 % C.I. (±)

373238253310981631

103353214

90 % C.I. (±)

677555597777176

2138

254493535

Source:SurveyofIncomeandProgramParticipationcov-ering calendar year 2012, along with federal budget data.Unlikeeveryothertableinthisstudy,thecostsshownhereare averaged over only the households that are using the welfareprograminquestion.Inotherwords,nozeroesareincluded in the averages.Sample sizes vary depending on the program.Householdsareclassifiedbythenativityofthehouseholdhead. C.I. = confidence interval.

Page 10: The Cost of Welfare Use By Immigrant and Native Householdstakes place. To address the undercount problem, CIS used a more complex dataset called the Survey of Income The Cost of Welfare

10

Center for Immigration Studies

AppendixTodeterminewhichhouseholdsusewhichwelfareprogramsintheSIPP,thisstudyfollowsalmosttheexactmethodologyofthepreviousCISreport,“WelfareUsebyImmigrantandNativeHouseholds”.TheoneexceptionisthattheaggregateCashcategorynowrefersexclusivelytoSSIandTANF.Inthepreviousstudy,Cashalsoincludedseveralmiscellaneousprogramssuchasstategeneralassistance,veterans’compensation,and,inthecrypticwordsoftheSIPPdocumentation,“otherwel-fare”.21 Because these smaller programs are too vaguely defined to locate in budget documents, the Cash category is now limitedtothetwolargestcashassistanceprograms,SSIandTANF.

Thecontributionofthisstudyistoestimatethedollarcostsassociatedwiththewelfareuseratespreviouslyreported.Thatprocessinvolvestwomajorsteps.First,someprogramcostsintheSIPPmustbeimputedbecausetheyarenotmeasureddirectlyinthesurvey.Second,duetomeasurementerrorinthesurveyresponsesandimputations,theSIPPcostsmustbeadjusted to reflect the full cost of each welfare program as reported in the government’s budget.

Program Costs in the SIPP. TheSIPPprovidesdirectcostestimatesforsomebutnotallofthewelfareprogramsforwhichitmeasuresparticipation.DeterminingtheannualhouseholdcostofSSI,TANF,WIC,andfoodstampsissimplyamatterofsummingthemonthlyvaluesofthevariablesprovidedintheSIPP.22However,thecostoffreeorreducedschoollunch,Medicaid,andsubsidizedhousing(the“cost-unknownprograms”)mustbeimputedseparately.

RecallthattheASECsubstantial-ly undercounts program partici-pation, making it suboptimal for awelfare analysis.However, theASEC does provide direct costestimates for the cost-unknown programs using imputation pro-cedures developed by the Cen-sus Bureau. The approach ofthis study is to apply the same imputation to the SIPP’s cost-unknown programs as the Cen-susBureauuseswiththeASEC.Theaimistoproducea“bestofboth worlds” dataset that com-bines the more accurate partici-pationratesoftheSIPPwiththemore complete cost estimates of theASEC.TableA1outlinestheprocess.

ForitsASECestimates,theCensusBureaudeterminedthattheaverageannualcostoffreelunchforonechildin2012was$507,whiletheannualcostofreducedlunchwas$440.TheSIPPdistinguishesbetweenfreeandreducedlunchreceiptatthehousehold level, so determining the monthlycostintheSIPP—rememberthattheSIPPmeasureswelfareusebymonth—issimplyamatterofdividingeither$507or$440by12,thenmultiplyingbythenumberofeligiblechildreninthehouseholdeachmonth.Theresultingmonthlyestimatesaresummedtoproduceanannualhouseholdcost.

Medicaidismorecomplicated.TheBureauidentifiesfour“riskclasses”—children(belowage21),non-elderlyadults,se-niors(age65andover),andthedisabled.23 It thenassignseachclassanaveragecostbystate.Thatgenerates4*50=200uniquecostestimates.Inaddition,MedicaidpaysthecostofMedicarepremiumsforpeopleenrolledinbothprograms.TheBureauaddstheannualMedicarepremium,whichwas$1,199in2012,tothetotalcostofMedicaidfordualenrollees.24ToperformtheimputationintheSIPP,thefullmatrixofcostestimateswasgeneratedintheASEC,thenmergedintotheSIPPusingthesameidentifyingvariablesofage,disabilitystatus,state,andMedicarecoverage.25Aswithschoollunch,theannualASECcostestimatesaredividedby12toproducemonthlySIPPcosts,thensummedtoproduceanannualfigure.

TheoriginalCISwelfarestudyseparatelyclassifiedpublichousingandsubsidizedrent,buttheASECcollapsesthecostofhousingassistanceintoonecategory.TheASECimputationisbasedonthreegroupsoffamilyincome(below$6,000,be-

Program

AnyWelfareCash

SSITANF

FoodSchool LunchWICSNAP

MedicaidHousing

PublicSubsidized

Table A1. Cost Calculation Methods by Program Source

Sumofcash,food,Medicaid,andhousinglistedbelowSumofSSIandTANF

DirectlymeasuredintheSIPPDirectlymeasuredintheSIPP

Sumofschoollunch,WIC,andSNAPAnnualcostofbothfreeandreducedlunchderivedfromASECDirectlymeasuredintheSIPPDirectlymeasuredintheSIPP

Averagecostbyage,disabilitystatus,andstate;derivedfromASECAveragecostbyregion,income,andsize;derivedfromASEC

Not individually calculatedNot individually calculated

SIPP=SurveyofIncomeandProgramParticipation.ASEC=AnnualSocialandEconomicSupplementtotheCurrentPopulationSurvey.

Page 11: The Cost of Welfare Use By Immigrant and Native Householdstakes place. To address the undercount problem, CIS used a more complex dataset called the Survey of Income The Cost of Welfare

11

tween$6,000and$10,000,morethan$10,000),threebedroomcounts(one,two,threeormore),andfourregions(Northeast,Midwest,South,West).26GeneratingreliablebedroomcountsistoodifficultwiththeSIPP,sothisstudyusesthenumberofpeopleinthehousehold(one,two,threeormore).AftergeneratingthefullmatrixofcostestimatesatthehouseholdlevelintheASEC—basedonregion,householdincome,andthenumberofpeople—thecostsaremergedintotheSIPPusingthosesamethreehouseholdcharacteristics.UnlikeschoollunchandMedicaid,housingcostsintheASECareprovidedasamonthlycost,meaningnodivisionby12isrequiredwhentransferredtotheSIPP.

Adjustment to Reflect Budgetary Costs. Afterallthewel-fare costs in the SIPP are established, the next step is toadjustthosecoststoreflecttherealbudgetarytotals.TableA2 compares the total costs reported (or imputed) in theSIPPwiththetotalcostslistedinthefederalgovernment’sbudgetaryrecords.ForconsistencywiththeSIPP,thebud-getary numbers collected here generally reflect the cost of actual benefits, not administrative costs. In addition, the cost of nursinghome care is excluded from theMedicaidbudget,sincetheSIPPdoesnotcovertheinstitutionalized population.

Some minor inconsistencies are unavoidable. For example, the budgetary cost of housing refers only to federal pro-grams, although state and local programs are a small part of thehousingcostmeasuredintheSIPP.Trackingdownthecost of every non-federal housing program would be infea-sible. In addition, most budgetary figures are available only for the “fiscal year”, which runs from October through Sep-tember,whereastheSIPPcoversthe“calendaryear”ofJanuarythroughDecember.27ThismeansthattheSIPPtimeperiodincludesninemonthsofFY2012(JanuarythroughSeptember)andthreemonthsofFY2013(OctoberthroughDecember).Tocorrectforthemisalignment,thisstudy’sbudgetarycostsforthe2012calendaryeararecalculatedasaweightedaverageoffiscal-yearcosts:(9/12)*(FY2012)+(3/12)*(FY2013).

Thereareseveralreasonswhythetotalcostofagivenprograminsurveydatadoesnotmatchthetotalbudgetarycostoftheprogram.First,thecostofdirecttransferssuchasTANF,WIC,andfoodstampsaretypicallyunderreportedinsurveys.Second,thecostimputationsofschoollunch,Medicaid,andhousingareinherentlyinexact.Third,thebudgetaryfiguresmaynot refer to the exact same costs measured in the survey, as is the case with housing.

CostsaregenerallyundercountedintheSIPP,butthediscrepanciesvaryconsiderablyfromprogramtoprogram,andSSIandschool lunch are actually overestimated.28Thesediscrepanciesmayseemlikeathreattothestudy’svalidity.Remember,how-ever,thattheSIPPcostestimatesaremerelyatoolfordividinguptherealbudgetary costs between immigrants and natives. Asnotedinthemaintext,theassumptionhereisnotthatimmigrantscostexactlywhatisreportedintheSIPP;rather,itisthatthefractionofcostsattributedtoimmigrantsintheSIPPisthesameasthefractionoftherealbudgetarycostsconsumedbyimmigrants.AnybiasintheSIPPdataunrelatedtoimmigrationstatus—forexample,ifbothimmigrantsandnativesun-derreporttheirTANFincomebythesamepercentage—isirrelevantoncethecostsareadjustedtoreflectbudgetarytotals.

Table1demonstratesthattheSIPPhelpscorrecttheASEC’sbiastowardundercountingimmigrantwelfareparticipationvis-à-visnativeparticipation.AslongasthereportedandimputedcostsofthatparticipationintheSIPParereasonablyunbiasedwith respect to immigration status, then the final adjusted cost estimates will be good approximations of the true budgetary numbers.

Theadjustmentprocedureissimple.ThecostofeachSIPPhousehold’sparticipationinagivenprogramismultipliedbyanadjustment factor, which is the ratio of total budgetary spending on the program to the total spending on the same program reportedintheSIPP.Forexample,theSNAPadjustmentfactoris$75.0billion/$56.5billion=1.33.Inotherwords,thetotalbudgetarycostofSNAPis1.33times(or33percent)higherthanthetotalSNAPcostintheSIPP.ThecostofeachSIPPhousehold’sSNAPusageismultipliedby1.33tobringthetotalSIPPcostsuptothebudgetarylevel.ThisismathematicallyequivalenttocalculatingtheportionofSNAPcostsattributedtoimmigrantsornativesintheSIPP,thenmultiplyingthatpercentagebythetotalbudgetarycostofSNAP.

Program

SSITANFSchool LunchWICSNAPMedicaidHousing

Table A2. Adjustment Factors for Program Costs in 2012

SIPP Cost(billions)

60.55.6

12.63.7

56.5251.017.3

Budgetary Cost(billions)

56.28.99.84.7

75.0361.947.6

Adjustment Factor

0.931.600.781.261.331.442.75

SIPP=SurveyofIncomeandProgramParticipation.AdjustmentFactoristheratioofbudgetarycosttoSIPPcost.Budgetary costs are modified. See text for details.

Page 12: The Cost of Welfare Use By Immigrant and Native Householdstakes place. To address the undercount problem, CIS used a more complex dataset called the Survey of Income The Cost of Welfare

12

Center for Immigration Studies

Supplemental Tables

Program

AnyWelfareCash

SSITANF

FoodSchool LunchWICSNAP

MedicaidHousing

Sample SizeWeightedn(millions)

Program

AnyWelfareCash

SSITANF

FoodSchool LunchWICSNAP

MedicaidHousing

Sample SizeWeightedn(millions)

Table A3. Average Cost of Household Welfare Use in 2012, by Age

Cost

5,704352212140

1,47286

1251,2623,183697

1,64510.89

Cost

4,0031023766

1,084134223726

2,6661511841.36

Cost

6,617517367151

1,35018268

1,1004,244

5053,21617.59

Cost

7,152419252167

1,740338160

1,2434,793

2005703.57

Cost

4,78056349171

6928625

5813,169356

3,83919.65

Cost

5,856458342116

1,12323470

8193,923

3527714.51

Cost

4,19268664640

427279

3912,766

3137,01432.21

Cost

2,309328308

20202

92

1911,462317

6,36324.26

Cost

6,105844791

537689132

6454,0684258644.19

Cost

7,0261,5191,465

53605278

5714,082820591

2.53

29and Under

29and Under

Native Households

Immigrant Households

30-39

30-39

40-49

40-49

50-6465

and Over

50-6465

and Over

90 % C.I. (±)

460605036

1251115

113295113

90 % C.I. (±)

865763162

2702963

230637157

90 % C.I. (±)

45972583489127

7932978

90 % C.I. (±)

88912410574

2183422

19262494

90 % C.I. (±)

3266863175574

5122948

90 % C.I. (±)

69612611060

1742019

160475114

90 % C.I. (±)

24557578

3432

3217240

90 % C.I. (±)

15539387

1611

16108

35

90 % C.I. (±)

720162168

301141410

107508132

90 % C.I. (±)

986414393409114

585

584171

Source:SurveyofIncomeandProgramParticipationcoveringcalendaryear2012,alongwithfederalbudgetdata.Householdsareclassifiedbythenativityofthehouseholdhead.C.I. = confidence interval.

Page 13: The Cost of Welfare Use By Immigrant and Native Householdstakes place. To address the undercount problem, CIS used a more complex dataset called the Survey of Income The Cost of Welfare

13

Program

AnyWelfareCash

SSITANF

FoodSchool LunchWICSNAP

MedicaidHousing

Sample SizeWeightedn(millions)

Table A4. Average Cost of Immigrant Welfare Use, by Household Head’s Length of Residence in U.S.

Cost

7,53776667690

1,152162137852

4,99762277

0.41

Cost

5,57657353142

1,033161101770

3,5604106343.61

Cost

6,537798644154

1,264206101957

4,0943815303.06

Cost

6,601532408125

1,37026297

1,0104,243455378

2.13

Cost

6,22576667393

851148

50653

4,238370

1,1705.92

Less than5

Immigrants’ Residence in U.S. (years)

5-10 11-15 16-25More than

25

90 % C.I. (±)

2,8845345351214677777

4032,038546

90 % C.I. (±)

72315415227

1562219

140474120

90 % C.I. (±)

85330328387

1962925

169533109

90 % C.I. (±)

1,00916813980

2904033

262681189

90 % C.I. (±)

69114714145

1322112

11949679

Source: Survey of Income and Program Participation covering calendar year 2012, along with federal budgetdata. Householdsareclassifiedbythenativityofthehouseholdhead. C.I. = confidence interval. SomeimmigrantsaremissingbecausetherelevantSIPPquestionwasaskedaspartofaone-timemodule.

Page 14: The Cost of Welfare Use By Immigrant and Native Householdstakes place. To address the undercount problem, CIS used a more complex dataset called the Survey of Income The Cost of Welfare

14

Center for Immigration Studies

Program

AnyWelfareCash

SSITANF

FoodSchool LunchWICSNAP

MedicaidHousing

Sample SizeWeightedn(millions)

Program

AnyWelfareCash

SSITANF

FoodSchool LunchWICSNAP

MedicaidHousing

Sample SizeWeightedn(millions)

Table A5. Average Cost of Welfare Use in 2012 in Households with a Worker, by Education

Cost

10,1851,006870136

1,94021676

1,6496,5746658523.79

Cost

9,072666533133

1,964381181

1,4016,1862565853.18

Cost

4,88245037476

94310054

7893,178

3113,49716.99

Cost

6,895581428153

1,405258107

1,0404,563346560

3.25

Cost

3,44629625640

6498338

5282,305196

5,66828.98

Cost

4,565469456

1370613679

4903,2311595082.95

Cost

92085796

131201299

67231

5,35526.99

Cost

1,899201188

13288

5221

2151,310

1017744.26

Less Than High School

Less Than High School

Native Working Households

Immigrant Working Households

High School

High School

SomeCollege

SomeCollege

College DegreeOr More

College DegreeOr More

90 % C.I. (±)

1,18020319183

2262417

201795194

90 % C.I. (±)

86512711656

2203129

198678105

90 % C.I. (±)

3265753218098

72232

50

90 % C.I. (±)

69914814572

1832728

162518121

90 % C.I. (±)

2023839124664

43153

31

90 % C.I. (±)

83630129917

1652620

14857691

90 % C.I. (±)

11223224

2033

18919

90 % C.I. (±)

3628381139289

8625656

Source: Survey of Income andProgramParticipation covering calendar year 2012, alongwithfederal budget data.Householdsareclassifiedbythenativityofthehouseholdhead.C.I. = confidence interval. Worker status is determined by the January interview.

Page 15: The Cost of Welfare Use By Immigrant and Native Householdstakes place. To address the undercount problem, CIS used a more complex dataset called the Survey of Income The Cost of Welfare

15

Program

AnyWelfareCash

SSITANF

FoodSchool LunchWICSNAP

MedicaidHousing

Sample SizeWeightedn(millions)

Program

AnyWelfareCash

SSITANF

FoodSchool LunchWICSNAP

MedicaidHousing

Sample SizeWeightedn(millions)

Table A6. Average Cost of Welfare Use in 2012 in Households with Children, by Race

Cost

8,536664470194

1,694210100

1,3845,549629

6,23932.01

Cost

8,626533351182

1,894378171

1,3455,839360

1,3087.53

Cost

11,216975607368

2,315392148

1,7756,8861,0416013.93

Cost

10,233608342267

2,454528242

1,6836,8802916604.20

Cost

18,2561,6721,187485

3,676400174

3,10210,9511,9568884.93

Cost

9,22322817949

1,818260126

1,4316,4057721080.63

Cost

5,70436828187

1,12513073

9223,977234

4,49321.69

Cost

5,679689684

5322159

8155

4,6670

610.35

Cost

6,855407290117

1,20321785

9014,675570276

1.32

Cost

4,83746141942

91412758

7293,327136232

1.23

All

All

Native Households with Children

Immigrant Households with Children

Hispanic

Hispanic

Black

Black

White Asian

White Asian

90 % C.I. (±)

35754462676107

6925863

90 % C.I. (±)

688988153

1562117

14148593

90 % C.I. (±)

1,1522161841012362925

215779249

90 % C.I. (±)

90813110789

2423228

215664112

90 % C.I. (±)

1,2852271871122732824

255873266

90 % C.I. (±)

2,583157121

524605849

4281,931474

90 % C.I. (±)

3694541197797

6927841

90 % C.I. (±)

3,325681679

9170599

1432,633

0

90 % C.I. (±)

1,43620516267

2793326

254914307

90 % C.I. (±)

1,270219212

333282729

30683391

Source:SurveyofIncomeandProgramParticipationcoveringcalendaryear2012,alongwithfederalbudgetdata.Householdsareclassifiedbythenativityofthehouseholdhead.C.I. = confidence interval.TheBlack,White,andAsiancolumnsexcludeHispanics.“Other”racenotshown.Child status is determined by the January interview.

Page 16: The Cost of Welfare Use By Immigrant and Native Householdstakes place. To address the undercount problem, CIS used a more complex dataset called the Survey of Income The Cost of Welfare

16

Center for Immigration Studies

Program

AnyWelfareCash

SSITANF

FoodSchool LunchWICSNAP

MedicaidHousing

Sample SizeWeightedn(millions)

Program

AnyWelfareCash

SSITANF

FoodSchool LunchWICSNAP

MedicaidHousing

Sample SizeWeightedn(millions)

Table A7. Average Cost of Welfare Use in 2012 in Households in Poverty, by Race

Cost

14,3681,301989312

2,855209110

2,5368,3181,8952,93214.37

Cost

13,086988683305

2,870372183

2,3148,0771,152612

3.33

Cost

19,9441,8741,221653

4,182434181

3,56810,8523,0362601.58

Cost

13,386818367451

3,438537268

2,6338,4486823221.94

Cost

20,8091,9741,461

5134,070

302150

3,61811,1823,582694

3.50

Cost

13,79663657165

2,43520891

2,1369,1921,534

520.30

Cost

10,556867721146

2,10513081

1,8936,590994

1,8138.52

Cost

11,681684684

067390

0582

9,1591,166

180.10

Cost

12,4411,3161,123193

1,95613664

1,7556,9242,246

1330.60

Cost

12,3291,6291,615

132,041

9050

1,9027,1581,501

950.44

All

All

Native Households in Poverty

Immigrant Households in Poverty

Hispanic

Hispanic

Black

Black

White Asian

White Asian

90 % C.I. (±)

5911079052

1361411

124407149

90 % C.I. (±)

1,230193162116285

3131

261806224

90 % C.I. (±)

1,8734363303014516040

4111,172672

90 % C.I. (±)

1,1432131581903894647

353810247

90 % C.I. (±)

1,592284222133323

3030

2941,041381

90 % C.I. (±)

6,10244042490

8877154

8294,2501,121

90 % C.I. (±)

688999438

1601313

147502127

90 % C.I. (±)

5,077823823

049259

0496

3,9541,213

90 % C.I. (±)

2,4225234731534934734

4541,419570

90 % C.I. (±)

3,800703702

217184236

6882,408621

Source:SurveyofIncomeandProgramParticipationcoveringcalendaryear2012,alongwithfederalbudgetdata.Householdsareclassifiedbythenativityofthehouseholdhead.C.I. = confidence interval. TheBlack,White,andAsiancolumnsexcludeHispanics.“Other”racenotshown.PovertystatusisdeterminedbytheJanuaryinterview.

Page 17: The Cost of Welfare Use By Immigrant and Native Householdstakes place. To address the undercount problem, CIS used a more complex dataset called the Survey of Income The Cost of Welfare

17

End Notes1 StevenA.Camarota,“WelfareUsebyImmigrantandNativeHouseholds”, Center for Immigration Studies Backgrounder, September 2015.

2 National Research Council, The New Americans, Washington,D.C.:NationalAcademyPress,1997,p.308.

3 TheHeritageFoundation’s2013fiscalanalysismakesthesameassumption.RobertRectorandJasonRichwine,“TheFiscalCostofUnlawfulImmigrantsandAmnestytotheU.S.Taxpayer”,HeritageFoundationSpecialReport,May6,2013,p.47.

4 HereisalistofthecountriesgroupedundertheregioncategoriesinFigure2.Central America:Belize,Mexico,CostaRica,ElSalvador,Guatemala,Honduras,Nicaragua,Panama.Caribbean:Barbados,Cuba,Dominica,DominicanRepublic,Gre-nada,Haiti,Jamaica,TrinidadandTobago,WestIndies.South America:Argentina,Bolivia,Brazil,Chile,Colombia,Ecua-dor,Guyana,Peru,Uruguay,Venezuela.Europe:Austria,Belgium,France,Germany,Ireland,Holland/Netherlands,Norway,Sweden,Switzerland,UnitedKingdom,Albania,Bulgaria,Greece,Hungary,Italy,Poland,Portugal,Romania,Spain,CzechRepublic,BosniaandHerzegovina,Croatia,Lithuania,Belarus,Russia,Ukraine,USSR.East Asia:China,HongKong,Ja-pan,Korea,SouthKorea,Taiwan.South Asia:Afghanistan,Bangladesh,India,Iran,Nepal,Pakistan,SriLanka,Uzbekistan.Africa:Egypt,EquatorialGuinea,Ethiopia,Eritrea,Ghana,Kenya,Liberia,Morocco,Nigeria,SierraLeone,Somalia,SouthAfrica,Sudan.

5 StevenA.Camarota, “WelfareUsebyLegal and Illegal ImmigrantHouseholds”, Center for Immigration Studies Back-grounder, September 2015, pp. 2-3.

6 Camarota, “WelfareUsebyImmigrantandNativeHouseholds”,Appendix.

7 See,forexample,ShikhaDalmia,“Heritage’sUpdatedStudyontheWelfareCostsofImmigrants:GarbageIn,GarbageOut”, Reason,May7,2013.

8 Camarota,“WelfareUsebyImmigrantandNativeHouseholds”,Appendix.

9 Ibid., pp. 22-23.

10 National Research Council, The New Americans,pp.284-289.The$2,200figurereferstoCalifornia,whichtheauthorsdescribe as likely to be an upper bound estimate.

11 RectorandRichwine,“TheFiscalCostofUnlawfulImmigrantsandAmnestytotheU.S.Taxpayer”,Table6.Bothtypesofhouseholds are in deficit because the government spends more than it takes in.

12 See, for example, Will Wilkinson, “WelfareandAmnesty”, The Economist,May8,2013.

13 ThemistakenbeliefthatimmigrationgenerateslargeindirectfiscalbenefitsseemstostemfromconfusingGDPgrowthwithgainsfornatives.Thelatterisjustatinyproportionoftheformer.Foramorein-depthdiscussion,seeJasonRichwine,“Mostof theGains fromImmigrationGotoImmigrantsThemselves–Not toNatives”, Center for Immigration Studies, February10,2016.

14 George Borjas, “ImmigrationandtheAmericanWorker”, Center for Immigration Studies Backgrounder,April2013.

15 See,forexample,GeorgeJ.Borjas,JeffreyGrogger,andGordonH.Hanson,“ImmigrationandtheEconomicStatusofAfrican-AmericanMen”, Economica,Vol.73(2010),pp.255-282.

16 RobertD.Putnam,“E Pluribus Unum:DiversityandCommunityintheTwenty-firstCentury”,Scandinavian Political Stud-ies,Vol.30,No.2(June2007),pp.137-174.

17 Camarota,“WelfareUsebyImmigrantandNativeHouseholds”.

Page 18: The Cost of Welfare Use By Immigrant and Native Householdstakes place. To address the undercount problem, CIS used a more complex dataset called the Survey of Income The Cost of Welfare

18

Center for Immigration Studies

18 Ibid.,Appendix.

19 Means-testedanti-povertyspendinghastrendedineluctablyupward.Eventhefamous“cuts”in1981and1996werejustminor speed bumps on the road to higher spending. See Jason Richwine, “Crying Wolf Over Spending Cuts”,TheRichwineArchive,January18,2014.

20 “HowMass(Legal)ImmigrationDoomsaConservativeRepublicanParty”,EagleForum,firstpublishedJanuary2014.

21 PersonalcommunicationwithShelleyIrvingattheCensusBureau.

22 OnereportedvalueforannualSSIreceiptisanimplausiblyhigh$69,934.Itappearstobeamis-code,asthenext-highestvalueisjust$37,281.Theimplausiblevaluehasbeentop-codedto$37,281forthisstudy. 23 AdisabledpersonintheASECisunderage65andhasatleastoneofthefollowingadditionalcharacteristics:SSIincome,Social Security income, or not working due to illness.

24 PersonalcommunicationwithJessicaSemegaattheCensusBureau.

25 Unfortunately,theSIPPdoesnotidentifyMedicarerecipientswhoareunder15yearsold,eventhoughchildrenwithkid-neydiseasecouldbeontheprogram.Therefore,thecostofMedicaidneverincludesanadditional$1,199Medicarepremiumfor respondents under 15 in this study.

26 PaulD.Johnson,TrudiRenwick,andKathleenShort,“EstimatingTheValueofFederalHousingAssistancefortheSupple-mentalPovertyMeasure”,U.S.CensusBureauWorkingPaper,December2010,pp.6-7.

27 TheSIPPpanelusedinthisstudystartedin2008,soitispossibletocalculateSIPPwelfarecostsoverthefiscal-yearperiodofOctober2011toSeptember2012ratherthanoverthecalendar-yearperiod.ThetroubleisthattheCensusBureaupro-videslongitudinalweightsonlyforcalendaryears.Withoutthoseweights,theSIPPwouldbeunrepresentativeofthegeneralpopulation.

28 OnereasonSSIisoverestimatedintheSIPPmaybethatrespondentsconfuseitwithdisabilityinsurance(DI)payments.BothSSIandDIareadministeredbytheSocialSecurityAdministration,butDIisasocialinsuranceprogramopenonlytoworkers who pay into the system, while SSI is a means-tested transfer payment.