Methodology

Probably the most important study on the fiscal effects of immigration was conducted by the National Research Council (NRC) in 1997.¹ Our analysis relies heavily on the approach used in the NRC study as the basis for estimating the fiscal impact of immigration. The NRC actually reported two different estimates for the fiscal impact of immigration, a household-level analysis of the current fiscal impact and an intergenerational analysis looking at immigrants and their descendents over a 300-year period. Our analysis primarily follows the example of the NRC's household level analysis because we are interested in estimating the current fiscal impact of illegals on the federal budget. However, we also report separate estimates for immigrants by education level as was done in the NRC's intergenerational analysis.

Another important study was conducted in 1998 by the Urban Institute. That study only estimated tax payments in New York State, including some federal taxes. But unlike the NRC study, the Urban Institute New York study estimated tax payments for legal and illegal immigrants separately, though it did not consider service use. This is one of the only studies today that has examined tax payments by illegal aliens, and so we rely on some of that study's approach as well.² Other important studies that have examined the fiscal impact of immigration include a 2001 study of Florida, a 1997 study of New Jersey (which was included in the NRC study), and one in 1994 by the Center for Immigration Studies.³

Almost all fiscal studies of immigration attempt to measure the taxes paid by immigrant households and the services they use. This study follows the same approach. We also make the same implicit assumption of almost all fiscal studies, including the NRC's: that if immigrants create a fiscal deficit, then taxes simply rise to cover the added expenses while services remain the same rather than taxes staying the same and services being reduced. Whether natives have to pay more to retain the same level of services or receive less in services for the same price, the outcome is still bad for them. Conversely, if illegal immigrants pay more in taxes than they use in services then this would be a clear benefit for natives because they could receive the same level of services but pay less in taxes.

Data Source and General Principles
Data Source. This report relies on the March 2003 Current Population Survey (CPS) collected by the U.S. Census Bureau. The March data, also called the Annual Social and Economic Supplement, includes an extra-large sample of minorities and is considered one of the best sources of information on the foreign-born.⁴ The foreign-born are defined as persons living in the United States who were not U.S. citizens at birth.⁵ For the purposes of this report, foreign-born and immigrant are used synonymously. The Survey includes most legal immigrants and is thought to capture roughly 90 percent of the illegal alien population. We use the term illegal alien or illegal immigrant to mean those who responded to the survey who are in the United States without authorization. All other foreign-born persons are referred to as legal immigrants, including those with Permanent Residence, those who are naturalized American citizens, and those living in the United States on long-term temporary visas, mainly guest workers and foreign students.

The CPS asks respondents about their income and program use in the calendar year prior to the Survey, so all fiscal estimates in the study are for 2002.⁶ Almost all past research on the fiscal impact of immigrants has relied on CPS or Decennial Census data, including the NRC and the Urban Institute studies. Information about actual taxes collected by the federal government by source comes from the Office of Management and Budget.⁷ Information on actual federal expenditures on means-tested programs comes from the Congressional Research Service.⁸ Information on other expenditures comes from a variety of government publications.⁹

Identifying Illegal Aliens. The CPS does not ask the foreign-born if they are legal residents of the United States. However, the Urban Institute, the former INS, and the Census Bureau have used socio-demographic characteristics in the data to estimate the size of the illegal population. To determine who are legal and illegal immigrants in the survey, this report uses citizenship status, year of arrival in the United States, age, country of birth, educational attainment, sex, receipt of welfare programs, receipt of Social Security, veteran status, and marital status. We use these variables to assign probabilities to each respondent. Those individuals who have a cumulative probability of one or higher are assumed to be illegal aliens. The probabilities are assigned so that both the total number of illegal aliens and the characteristics of the illegal population closely match other research in the field, particularly the estimates developed by the Urban Institute.

This method is based on some well-established facts about the characteristics of the illegal population. For example, it is well known that illegals are disproportionately male, unmarried, under age 40, have few years of schooling, etc. Thus, we assign probabilities to these and other factors in order to select the likely illegal population. In some cases we assume that there is no probability that an individual is an illegal alien. If an individual reports that he is U.S.-born or a naturalized citizen of the United States, then he is assumed not to be an illegal alien. Someone who reports that he is veteran or receives veteran benefits is also assumed not to be an illegal alien. Those individuals who report that they personally receive Social Security benefits, cash assistance under Temporary Assistance to Needy Families (TANF), Supplemental Security benefits (SSI), or who are enrolled in Medicaid are also assumed not to be illegal aliens. However, other members of a household headed by an illegal alien can receive these programs, mostly the U.S.-born children of illegals. It is worth noting that our findings show that only a tiny fraction of households headed by illegals receive cash welfare programs or Social Security benefits. However, a large share of children in illegal alien households use the school lunch program or are enrolled in Medicaid. Our methodology allows for such a possibility.

We estimate that there were 8.7 million illegal aliens in the March 2003 CPS. It must be remembered that this estimate only includes illegal aliens captured by the March CPS, not those missed by the survey.¹⁰ By design this estimate is very similar to those prepared by the Census Bureau, the former Immigration and Naturalization Service (INS), and the Urban Institute.¹¹ Although it should be obvious that there is no definitive means of determining whether a respondent in the survey is an illegal alien, the findings in this study are consistent with previous research. For example, the Urban Institute estimated that in 2002 Mexicans accounted for 57 percent of the illegal population; our method finds 58 percent in 2003. Using 2003 data, we estimate that 88 percent of illegals arrived after 1990; the Urban Institute estimated 85 percent using 2002 data.¹² Our results also produce estimates that are similar in other areas, such as age and workforce participation.

Unit of Analysis. We divide households between those headed by illegal aliens and all others. In reference to its fiscal estimates, the NRC states, "Since the household is the primary unit through which public services are consumed and taxes paid, it is the most appropriate unit as a general rule and is recommended for static analysis."¹³ Because our study is also focused on "static analysis," or current fiscal effects of illegal aliens, we also examine taxes paid and services used by households based on the nativity and legal status of the household head. Like the NRC study, we define households as all persons living together who are related. Persons living with individuals to whom they are unrelated or who live alone are considered their own household. This definition could also be referred to as a "family," but following the NRC's example, we call it a household.

In their study of New Jersey, Deborah Garvey and Thomas Espenshade also used households as the unit of analysis because "households come closer to approximating a functioning socioeconomic unit of mutual exchange and support."¹⁴ Another reason for using households is that Census Bureau surveys are collected by household, making households the most appropriate use of the survey. Even so, it must be remembered that grouping by household, even the modified definition of household used in the NRC study and this report, means that many children born in the United States to illegal aliens are included in illegal alien-headed households even though these children are U.S. citizens by virtue of being born here. This seems perfectly reasonable since the presence of these children in the United States is a direct result of their parents having been allowed to enter and remain in the United States. Thus, counting services used by these children allows for a full accounting of the costs of illegal immigration.

Marginal vs. Average Costs. Like the NRC, we assume that average costs equal marginal costs. That is, an additional person or household using a program or service costs the same as those already using the service. This, of course, is not always the case. For example, the addition of a few new students to a half empty school costs relatively little because the school is already built and so no additional funding is needed for school construction. In such a situation the marginal costs of the new students are much less than average costs for the students already in the school. On the other hand, an additional group of students added to an already overcrowded school may require a whole new school to be built, thus making the marginal costs of the additional students much greater than the average cost. The NRC and others assume that marginal and average costs are equal and that these two tendencies should balance each other out over time.

Private vs. Public Goods. Some goods provided by government are pure public goods; that is, everyone living in the country benefits from receiving them. At the same time, the cost of providing them does not rise as the number of people living in the country increases. The most important example of this type of program is national defense. Like the NRC study, we assume that defense is a pure public good at the federal level, therefore no costs for providing defense are assigned to immigrant households. We further assume that, with the exception of means-tested programs for veterans, which illegals cannot use, all non-means-tested veterans programs are also pure public good. This means that the nearly $388 billion spent on defense and veterans programs in 2002 is assigned only to native households.

Interest on the National Debt and Federal Deficits. Following the example of the NRC, we do not include interest payments on the national debt in our calculations of costs because it is impossible to determine what share of the debt was incurred due to illegals. Obviously, illegal aliens who just arrived in the country have not contributed at all to past deficits or to the cumulative total of those deficits -- the national debt. On the other hand, if immigrant households created a net fiscal burden in past years then they might account for a disproportionate share of the current national debt. Thus, like most previous studies, including the NRC, we do not count interest on the national debt as a cost for either immigrant or native households. This means that the study measures only the fiscal impact of illegal alien households on what the NRC calls the "primary" budget. The primary budget is comprised of all tax payments and all federal expenditures other than interest payments on the national debt.

Because debt interest payments ($170.95 billion) were slightly larger in size than the federal deficit ($157.80 billion) in 2002, the primary budget had a $13.15 billion surplus in 2002. However, we exclude from our analysis the $23.7 billion that the Federal Reserve earned from interest on federal reserve deposits, which come from investing in U.S. government securities. These monies are in effect the federal government paying itself and, although these monies are officially considered revenue, we do not count them as such in this report because we do not include the interest the federal government pays on the national debt as an expense. As a consequence, there was a $10.5 billion deficit ($85 per household) in the primary budget when interest earned on federal reserve deposits is excluded.

Economic Impact of Illegal Aliens. Like most studies of this kind, including the NRC's, ours does not consider how illegal immigration or immigration more generally might affect public coffers indirectly by its impact on the economy. There simply is no consensus on the economic impact of immigration. To the extent that the issue has been studied, the impact on the nation's economy is generally thought to be trivial relative to the size of the economy. In addition to its fiscal estimates, the National Research Council estimated that immigration created a net economic benefit to natives of between $1 billion to $10 billion in the mid-1990s, or an amount equal to one or two-tenths of 1 percent of the nation's economy at that time. And these figures are for all immigrants, not specially for the one-fourth of the foreign-born who are illegal aliens. Moreover, the same study found that immigration reduced the wages of native-born workers who lack a high school education by about 5 percent. Not only would this reduction lower the tax payments of unskilled natives, but it would almost certainly result in higher use of means-tested programs by these workers, who roughly correspond to the poorest 10 percent of the workforce. Thus it is not clear that the economic impact of illegal immigration would have even a tiny net positive effect on the public coffers. A more recent study from the National Bureau of Economic Research suggests that immigration, legal and illegal, has a decidedly negative impact on the income of all Americans.¹⁵ If that study is correct and there is a net loss for native-born Americans, then the tax payments of immigrants are much lower, while their use of services is higher as result of their lower incomes. Because the actual economic impact is probably modest relative to the overall size of the U.S. economy and there is little agreement on whether the effect is positive or negative, we follow the example of almost all other studies and focus on the direct taxes and services illegals pay and use.

Estimated Tax Payments
All studies of this kind involve estimating payroll and other taxes paid by households based on their characteristics: primarily income, number of dependents, and home ownership. In this study, all taxes collected by the federal government are assigned to households. There is general agreement that excise taxes and payroll taxes, including those paid by employers such as unemployment, are ultimately borne by households. However, there is some debate about who actually pays coporate income tax--consumers or owners of capital. In this study we follow the example of the Office of Management and Budget and assume that owners of capital pay corporate income tax. All tax payments are adjusted to reflect actual total taxes collected by the federal government from each source for the year.¹⁶ We assign all taxes to U.S. residents and ignore the small share of persons living outside of the United States who pay federal taxes.

Payroll Taxes. There are four main payroll taxes collected by the federal government: income, Social Security, Medicare, and Unemployment Compensation. The Current Population Survey contains income tax and Social Security tax liabilities calculated by the Census Bureau for all tax-paying units in the Survey. In short, the Census Bureau uses data from the American Housing Survey, the Income Survey Development Program, and the Internal Revenue Service and combines this information with CPS data to create simulations of tax liabilities for all persons reporting income.¹⁷ We use the tax liability estimates from the Census Bureau to calculate federal income and Social Security tax liability for each household.¹⁸ We calculate both Medicare and Unemployment tax as a share of earning. Again, both the employee and employer share are assigned to households because previous research indicates that even the employer share of payroll taxes is ultimately borne by workers.

Corporate Income Tax. There is some debate about who actually pays corporate income taxes--owners of capital or consumers. The Office of Management and Budget in its estimates of the tax burden on American households distributes corporate income taxes based on each household's share of capital income, and we do the same in this study. We calculate corporate income tax as a share of interest and dividend income as reported in the CPS.

Excise and Estate Taxes. In addition to payroll taxes, the federal government collects taxes on a number of goods, mainly tobacco, alcohol, transportation fuels, and tires. We follow the NRC study's approach and allocate tobacco and alcohol taxes based on the number of people of drinking and smoking age in each household. Excise taxes, mainly the telephone tax and those collected for the highway and airport trust funds, are allocated as a share of household income. All tariffs collected on foreign goods are also assumed to be borne by consumers and are therefore allocated as a share of household income. Like the NRC study, we also allocated estate taxes to native households and to immigrant households that have been in the country for more than 20 years. Since the methodology we use to identify illegal aliens assumes that the vast majority of illegal aliens have been in the country for less than 20 years, the contributions of illegal aliens are very small.

Tax Compliance. While illegal aliens are assumed to pay their share of non-payroll taxes, payroll taxes collected by employers or income and other taxes paid by the self employed are a different matter. There is some research to indicate that only about half of illegal aliens are "paid on the books."¹⁹ That is, income and other taxes are withheld from their pay. In their study of New York State, Jeffrey Passel and Rebecca Clark assumed 60 percent compliance. In their study of New Jersey, Clark and Zimmermann assumed a 56 percent compliance rate.²⁰ New York and New Jersey are somewhat unusual because a smaller fraction of each state's illegal population is employed in agriculture than is true nationally, a sector where being paid off the books is very common. However, we follow the same basic approach and assume that 55 percent of illegal aliens are paid on the books. We implement this by reducing the Income Tax and Social Security, Medicare, and Unemployment Tax of illegal households to 55 percent of their estimated tax liability.

Assigning Costs by Household
The CPS asks respondents a host of questions about their use of means-tested and non-means-tested federal programs. Like the NRC study and virtually all other studies on this topic, we use these results to estimate immigrant and native use of federal programs. Because the CPS began asking many more questions about use of public services in 2001, we are able to make specific estimates for a larger number of programs than was possible at the time of the NRC study, which relied on the 1995 CPS and 1990 Census. It should be remembered that the March 2003 CPS, the data source used in this study, asks questions about use of federal programs in the calendar year prior to the Survey, therefore all estimates reported here are for 2002. In that year expenditures in the primary federal budget totaled $1.84 trillion.

Social Insurance. We use the results in the CPS to estimate household receipt of Social Security, Medicare, and Unemployment Compensation. We assume that illegals receive no federal disability because these payments primarily go to federal workers or coal miners with black lung. For Social Security, we use responses in the CPS on the amount received. The same is true of Unemployment Compensation. The Survey also asks about receipt of Medicare. (Contrary to the common perception, persons under age 65 can qualify for Medicare, mainly those with end-stage renal disease.) For Medicare, we assigned average costs of the program to those who indicated they received it. As already indicated, we assume that persons getting Social Security cannot be illegal aliens. However, in a few cases members of their family can use these programs. As a result, costs for these two programs for illegals are extremely low relative to the rest of the population, but not zero.

Food Stamps and Cash Assistance. The CPS asks respondents the size of the payment they receive from the following programs: Supplemental Security Income, Temporary Assistance to Needy Families (TANF), cash assistance for low-income veterans, food stamps, low-income energy assistance, and higher education assistance. We use the payments respondents report to directly estimate average cost by household. Because food stamp values are reported in the CPS using the standard definition of household, and because we use a modified definition of household, food stamp values received by households with multiple families are divided based on the size of the each family in the household.

Other Non-Cash Means-Tested Programs. The Survey also asks about receipt of the Women Infants and Children program (WIC) and Free School Lunch program²¹ and whether someone lives in public housing or receives a rent subsidy. For WIC, public housing, and rent subsidies we assigned average costs of each program to households receiving it. For Medicaid, the Census provides estimates by disability status for all beneficiaries in the CPS. Again, it should be remembered that only federal costs of the program are considered in this study. For social services provided by the TANF program, we assigned costs evenly to households based on receipt of TANF. For government subsidized daycare, which relatively few illegals use, we assign average costs for households that indicated in the CPS that they receive child care services. College students in non-illegal households are assumed to be getting Stafford student loans if the household income is less than $75,000 a year. For illegal households, student loans are assumed only if the college student himself is an American citizen. Thus, while there are a few college students from illegal alien households receiving student loans, the number is extremely small. Illegal alien households are assumed to impose no costs on programs designed only for refugees. The same is true for programs for low-income veterans because our methodology assumes that all persons who indicate they are veterans cannot be illegal aliens.

The Earned Income Tax Credit. Based on income and other family characteristics, the Census Bureau provides estimated payments for the Earned Income Tax Credit (EITC), which it includes in the public use files of the CPS. We assume that natives and legal immigrants receive their EITC payments, but for illegal aliens we assume that only the 55 percent who are paid "on the books" can receive the program. Furthermore, we assume that one-fourth of those who are paid on the books and who also qualify for the program actually get it. To implement this, we reduce the Census Bureau's estimated EITC payments by 86 percent for illegal alien households. It should be noted that, officially, a valid Social Security number is required to collect the EITC. However, a Treasury Department Inspector General for Tax Administration report found that a small number of illegals without valid Social Security numbers did receive the EITC.²² Using this approach we estimate that illegal households only account for 1.5 percent of the total costs of the EITC. However, if they received the payments they qualify for based on their income and number of dependents they would account for more than 10 percent of the program's total costs.²³

The ACTC. In addition to the EITC, there is the Additional Child Tax Credit (ACTC), also called the refundable portion of the Child Tax Credit, which pays out a total of $5 billion a year to low-income workers with children. To estimate the cost of this program, we assign benefits to all native and legal immigrant households that have earnings over $10,000 and no federal income tax liability, based on the number of children under age 16 in the household. For illegals, we further assume that only 40 percent of those households that qualify based on our analysis of the CPS actually get the ACTC. This is much higher than for the EITC because illegals are explicitly allowed by law to get the credit. Thus they can receive the credit even if they do not have a stolen identity and Social Security number. If they don't have a valid SSN, they can still get the credit by obtaining an Individual Taxpayer Identification Number (ITIN) from the IRS, which is not difficult, and filing a return using the ITIN. The 2004 report by the Treasury Department Inspector General for Tax Administration mentioned above found that, in 2001, 203,000 tax forms filed by illegals got a cash payment from the Child Tax Credit using an ITIN. These illegals received a total payment of $161 million in that year--even if the SSNs on their W-2 forms were not valid.

This amount indicates that about 30 percent of illegals we identified in the CPS receive benefits from the program. We adjust this up to 40 percent because the Inspector General's report did not attempt to estimate receipt of the program by illegals who use stolen identities and Social Security numbers to file their returns; only use of the ITIN was considered. This adjustment has the effect of increasing the total cost of the program for illegals to $216 million in 2002.

Primary and Secondary Education. We allocate almost all of the federal funds for public education by household based on the number of school-age children in each household. The only exceptions are programs designed to assist schools that have a large number of children whose parents are migrants and those funds that specifically go to children with limited English. For programs designed specifically to improve the education of low-income children whose parents work in agriculture--mainly the Chapter I Migration Education Program -- we assign costs based on whether the household head works in agriculture and the number of school age children in the household. To allocate federal funds for children with limited English, we use the 2000 Census to calculate the share of school-age children who reported that they spoke English less than very well. The 2000 Census showed that 29.2 percent of school-age children in immigrant households (legal and illegal) spoke English less than very well, compared to 2 percent of those in native households. We assume the same percentages existed in 2002 and assigned costs accordingly.

Uninsured. Research by Jack Hadley and John Holahan indicates that federal expenditures for the uninsured totaled $21.03 billion in 2002.²⁴ We exclude from this total the $3.98 billion estimated to have been spent by the Department of Veterans Affairs on those without health insurance. (Illegals cannot receive health care from the VA and veteran benefits of this kind are treated as a pure public good in this analysis, so they are assigned only to native households). This means that $17.05 billion was spent by the federal government on the uninsured, not counting treatment from the VA. We then allocate these costs to households based on the number of uninsured persons per household.

INS, Federal Prisons, and Courts. We assign the net costs of the Immigration and Naturalization Service (expenditures minus fees the service collected), by household based on the distribution of immigrants (legal or illegal) who indicated they arrived after 1980.²⁵ It should be noted that not all of the net costs of the INS are attributable to immigrant-headed households because some post-1980 immigrants live in native households. We estimate that about 8 percent of post-1980 immigrants live in households headed by natives and as such, native households account for 8 percent of the net costs of the INS.²⁶ We do not include costs for running the immigration functions of the Department of State because this system is paid for by fees. As for the the federal prison system, it keeps track of whether inmates are citizens of the United States or not. In 2002, nearly 29 percent, or 39,000 inmates in the federal prison system were non-citizens. Based on prior research, we estimated that 59 percent of this total are illegal aliens.²⁷ This translates into 17 percent of the federal prison population and thus 17 percent of the $4.1 billion prison budget can be attributed to illegal alien households. For the cost of administering the federal court system, we again assume that 17 percent of the costs are attributable to illegal households. This estimate is probably too low because non-citizens in 2001, the last year for which data is available, accounted for 38 percent of those arrested by federal agencies and 34 percent of those actually convicted in federal courts.²⁸ This is significantly higher than the 29 percent of the prison population they represent. Nonetheless, we assign only 17 percent of the $4.7 billion federal court system budget to illegals in order to make our estimates more conservative.²⁹ As is the case for the federal prison system, when later in this report we estimate costs for households headed by legal immigrants we assume that 16 percent of the costs of the federal courts are due to households headed by legal immigrants (naturalized and unnaturalized).

All Other Expenditures. Using the results from the methodology described above, we are able to estimate almost $1.1 trillion in federal spending, or 58 percent of the primary budget. In addition to specific programs, we account for $388.1 billion, or another 21 percent of the primary 2002 federal budget, that went to spending on defense and non-means-tested veterans programs. As already noted, only native households are assigned costs for defense and veterans programs because they are assumed to be pure public goods and as such legal and illegal immigrants impose no costs on these programs. The remaining $386 billion, or 21 percent of the primary budget, is assigned to all households equally. This totals $3,115 per household. These expenditures include highway and infrastructure maintenance, parts of the criminal justice system not accounted for already, subsidies to business, state aid, and all other services provided by the federal government. This is the same approach used by the NRC and almost all studies of this kind. It should be noted that allocating costs equally to all households may tend to underestimate the costs of illegals because illegal households are 17 percent larger on average than other households. All other things being equal, more people per household should mean higher average expenses. But we ignore this and assign costs equally to all households.

Adjustment for Under-Reporting in the CPS
Under-Reporting of Program Use. It is well established that respondents tend to understate both their income and use of social services in the CPS. This problem is well known by the Census Bureau and has been studied for some time.³⁰ To correct for this problem, we adjust all social programs to reflect actual federal expenditure. This is based on the assumption that immigrants and natives are equally likely to under-report their use of social services. The NRC study also seems to have controlled for this problem.³¹ Adjustments of this kind may tend to understate program use by immigrants because they may be more reluctant to report use of means-tested programs in a government survey than other members of society out of a fear that this might constitute grounds for deportation. But we ignore this problem and assume that under-reporting rates are the same for all persons. As mentioned earlier, controlling all costs to actual expenditures has the added advantage of allowing us to estimate the costs of illegals not counted in the CPS. In effect, those who are counted are assigned costs for those who are missed by the survey. The same is true for non-illegal aliens who are missed by the survey. This means our cost estimates are for all illegals, even those the survey misses.

Under-Reporting of Income and Taxes. For tax payments, we adjust upward the income tax estimates calculated by the Census Bureau to reflect actual tax receipts. However, following the example of the Urban Institute in its study of New York State, we do so only for households with incomes of over $200,000 a year. This adjustment is based on the assumption that it is high-income households who under-report their income. While the Urban Institute first adjusted income for those with high income and then recalculated taxes, by adjusting tax receipts our approach has the same effect. For Social Security, we adjust payments for all taxpayers, not just those with high incomes, because only the first $84,900 of earnings was subject to the tax in 2002. Thus, under-reporting by high income earners does not matter. As is the case with costs, controlling all tax payments to the actual tax totals received by the federal government allows us to estimate the tax payments of illegals not counted in the CPS. In effect, those who are counted are assigned costs for those who are missed by the survey. The same is true for non-illegal aliens who are missed by the survey. This means that our tax estimates for all illegal and non-illegal households reflect the total taxes these populations pay, even those not counted by survey.

Return to Table of Contents

Continue to Findings

¹ The National Research Council report entitled, "The New Americans: Economic, Demographic, and Fiscal Effects of Immigration" can be ordered or read online at http://www.nap.edu/books/0309063566/html

²The Urban Institute study is entitled "Immigrants in New York: Their Legal Status, Incomes, and Taxes," by Jeffrey S. Passel and Rebecca L. Clark. A summary of the report can be found at http://www.urban.org/url.cfm?ID=407432

³The Florida study was conducted by the University of Florida and is entitled, "Facts about Immigration and Asking Six Big Questions for Florida and Miami-Dade." Information about the study can be found at http://www.bebr.ufl.edu/Articles/bosmig.pdf. The 1997 study of New Jersey is entitled "State and Local Fiscal Impacts of New Jersey's Immigrant and Native Households," and can be found in the edited volume "Keys to Successful Immigration: Implications of the New Jersey Experience," edited by Thomas J. Espenshade, Urban Institute Press. Washington DC. The Center for Immigration Studies study is entitled "The Costs of Immigration: Assessing a Conflicted Issue," by David Simcox, John Martin, and Rosemary Jenks.

⁴The survey is considered such an accurate source of information on the foreign-born because, unlike the decennial census, each household in the CPS receives an in-person interview from a Census Bureau employee. The 217,000 persons in the Survey, 23,000 of whom are foreign born, are weighted to reflect the actual size of the total U.S. population. However, it must be remembered that some percentage of the foreign born (especially illegal aliens) are missed by government surveys of this kind, thus the actual size of this population is almost certainly larger. The CPS data used in this study was provided by Unicon Research. www.unicon.com

⁵This includes naturalized American citizens, legal permanent residents (green card holders), illegal aliens, and people on long-term temporary visas such as students or guest workers, but not those born abroad of American parents.

⁶We supplement this data, where needed, with estimates derived from the 2000 Census because the Census contains several important pieces of information not found in the CPS, such as if a person speaks English well. We use this data to apportion the costs of providing for students who have limited English. The CPS, in contrast, does not ask a question about language.

⁷ Tables from the OMB showing taxes collected by source and year can be found at
http://www.whitehouse.gov/omb/budget/fy2004/hist.html

⁸ "Cash and Noncash Benefits for Persons with Limited Income: Eligibility Rules, Recipient and Expenditure Data, FY2000-FY2002," November 25, 2003. RL32233. Complied by Vee Burke.

⁹ Federal expenditures on primary and secondary education programs come from the Census Bureau report, "Federal Aid to States for Fiscal Year 2002," which can be found at http://www.census.gov/prod/2003pubs/fas02.pdf. Expenditures on Social Insurance Programs (Social Security, Medicare, and Unemployment Compensation) come from the OMB's web site, http://www.whitehouse.gov/omb. Figures for non-citizens in federal prisons come from the Federal Bureau of Prisons web site, http://www.bop.gov/fact0598.html#Population. The INS 2002 budget comes from the Department of Justice web site and can be found at: www.usdoj.gov/jmd/budgetsummary/btd/1975_2002/2002/html/page104-108.htm

¹⁰ Since all of the costs are controlled to the actual total taxes collected and total expenditures, our method has the effect of controlling for the undercount of illegal aliens, legal immigrants, and even natives.

¹¹ The INS report that estimated seven million illegals in 2000, with an annual increase of about 500,000, can be found at http://uscis.gov/graphics/shared/aboutus/statistics/Illegals.htm. The Census Bureau report with an estimate of eight million illegals in 2000 can be found at www.census.gov/dmd/www/ReportRec2.htm (Appendix A of Report 1 contains the estimates). The Urban Institute is the only organization to release figures for the size of the illegal population based on the CPS. Urban estimates that, in March of 2002, 8.3 million illegal aliens were counted in the CPS, with an additional one million being missed. Assuming continual growth in the CPS, there were between 8.6 and 8.8 million in the March 2003 CPS. Urban's estimates based on the March 2002 CPS can be found at http://www.urban.org/url.cfm?ID=1000587. Additional information was provided by Jeffery Passel of the Urban Institute in a May 24, 2004, telephone interview.

¹² The fact that our results match those from the Urban Institute is not an accident because the probabilities assigned to variables such as citizenship status, year of arrival in the United States, country of birth, or any of the other variables discussed above, are designed to match the results of previous research in this area.

¹³ "The New Americans," pp. 255-256.

¹⁴ "Keys to Successful Immigration: Implications of the New Jersey
Experience," p. 143.

¹⁵ The NBER paper, authored by David E. Weinstein and Donald R. Davis, can be found at www.columbia.edu/~drd28/Migration.pdf.

¹⁶Actual taxes collected by source can be found at OMB's web site,
http://www.whitehouse.gov/omb

¹⁷ A detailed discussion of how the Census Bureau estimates tax payment can be found in Census paper P60-186RD, "Measuring the Effect of Benefits and Taxes on Income and Poverty: 1992."

¹⁸ In the case of Social Security, estimated tax payments from the Census Bureau are roughly doubled to reflect employer contributions.

¹⁹ See David North and Marion F. Houstoun, "Characteristics and Role of Illegal Aliens in the U.S. Labor Market," 1976, Washington, D.C.: Linton and Co.; and Louis Rea and Richard Parker, "Illegal Immigration in San Diego Country: An Analysis of Costs and Revenues," 1992. Report to the California State Senate Special Committee on Border
Issues.

²⁰ "Immigrants in New York: Their Legal Status, Incomes, and Taxes;" "Undocumented Immigrants in New Jersey: Numbers, Impacts and Policies" in "Keys to Successful Immigration: Implications of the New Jersey Experience." Edited by Thomas Espenshade, 1997, Washington, D.C.: Urban Institute Press.

²¹ Although the CPS does not specifically ask about the breakfast program, we assign costs for this program to households based on use of the lunch program.

²² The Treasury Department Inspector General for Tax Administration report, # 2004-30-023, can be found at http://www.treas.gov/tigta/2004reports/200430023fr-redacted.html

²³ There are two ways an illegal alien could have a Social Security number that allows him to file a return and get the EITC: First, a stolen or otherwise aquired number and matching name from someone authorized to work in the United States. Second, some illegal aliens have been issued Social Security numbers. This can include illegal aliens who have an application pending with the immigration service, such as asylum applicants. Even thought the individual is illegally in the country and the application has only a small chance of ever being approved, some are still given work authorization. As IRS publication 596 states (page 6), Social Security numbers that are "Valid for Work only with INS authorization" can receive the EITC. The person need not be a legal resident of the United States. While it is unknown how many illegal immigrants have illegally acquired Social Security numbers, the number is likely to be significant. If roughly half of the 5.7 million illegals in the 2003 CPS who hold jobs are paid on the books, then this means that roughly three million illegals have provided employers with a SSN or perhaps an Individual Taxpayer Identification Number (ITIN) that was accepted. The ITIN is not supposed to be used by employers to pay payroll taxes, but this restriction may not be completely enforced.

²⁴ In a February 2003 study in Health Affairs, which can be found at http://www.healthaffairs.org, Hadley and Holahan estimated that the federal government spent $19.9 billion for the uninsured in 2001. An updated study for the Kaiser Family Foundation, which can be found at http://www.kff.org, estimated the figure was $23.5 billion in 2004. To estimate costs in 2002, we assume a constant rate of growth of 5.7 percent between 2001 and 2004, meaning that federal expenditures totaled $21.03 billion in 2002. Hadley and Holahan's two studies indicate that care provided directly by the federal government, such as the VA, increased at an annual rate of 2.4 percent. As a result, we estimate that veterans expenditures in 2002 were $3.98 billion.

²⁵ This is based on the assumption that the costs of administering the INS in 2002 were due to the presence of immigrants in the country either as legal immigrants who need a variety of services or as illegal aliens who are the subject of enforcement.

²⁶ We also use statistics for the Coast Guard interdiction program which can be found at http://www.uscg.mil/CG_2004_html/goals.html#migrant

²⁷ According to the Justice Department, non-citizens comprised 28.8 percent of the federal prison population in 2002. Figures can be found at http://www.bop.gov/fact0598.html#Citizenship. An unpublished paper by Rebecca Clark and Scott Anderson from the Urban Institute, entitled "Illegal Aliens in Federal, State, and Local Criminal Justice Systems," indicated that 57 percent of non-citizens in 1996 were illegal aliens and the share was increasing by 0.4 percentage points a year in the mid-1990s. Thus, by 2002, about 59 percent of non-citizens were likely illegal aliens. This means that 23,000 (22,978), or 17 percent, of the federal prison population were illegals. When later in this report we estimate costs for households headed by legal immigrants we assume that 11.8 percent (28.8 percent - 17 percent) of non-citizens in federal prisons are legal immigrants, and 4.5 percent of the federal prison population are naturalized U.S. citizens, which is their share of the nation's total population. This means that 16 percent of federal prisoners are legal immigrants (naturalized and unnaturalized).

²⁸ Information on those in federal courts can be found in the Compendium of Federal Justice Statistics: http://www.ojp.usdoj.gov/bjs/abstract/cfjs01.htm

²⁹ The costs of the federal court system can be found at
http://www.uscourts.gov/ttb/mar03ttb/budget.html

³⁰ See Appendix F in "Measuring the Effect of Benefits and Taxes on Income and Poverty: 1992" from the U.S. Census Bureau for a discussion of under-reporting of income and receipt of redistribution programs.

³¹ Footnote 32 on page 278 of the NRC report indicates that households that received redistribution programs like Social Security are identified in the CPS and then assigned the average benefit level. This would seem to control for under-reporting.