Immigrants in New York
Their Legal Status, Incomes, and TaxesPublication Date: April 01, 1998 Other Availability:  Printer-friendly summaryPermanent Link: http://www.urban.org/url.cfm?ID=407432 Share:      
The research was supported by grants from the New York Community Trust, the Emma Lazarus Fund, the Robert Sterling Clark Foundation, and the Heron Foundation. Additional support was provided by the Urban Institute's Program for Research on Immigration Policy, funded by The Ford Foundation and the Immigrant Policy Program funded by the Andrew W. Mellon Foundation.
Outline of ReportI. Executive SummaryIntroduction This report provides essential demographic and economic information on legal immigrants residing in New York State and addresses significant shortcomings in the existing data for immigrants and in analyses of fiscal impacts of legal immigrants. It focuses on four major issues: the size of the legal immigrant populations; the characteristics of legal status groups, including both legal and undocumented populations; the incomes and taxes paid by immigrant populations and natives; and the economic adaptation of immigrants and their descendants. The report is the first to provide population estimates of immigrants by legal status. It gives separate estimates of the number of naturalized citizens, legal permanent resident aliens ("green card" holders), refugees, legal nonimmigrants (such as diplomats, foreign students, and international business transfers), and undocumented aliens residing in New York. It also offers the first detailed estimates of the incomes and taxes paid by each of the five immigrant groups, as well as by natives residing in New York State. Finally, the report provides information on immigrants' adaptation to the United States— tracking how their income and tax contributions change as their time in the United States increases, and showing how the incomes and tax contributions of the U.S.-born adult offspring of immigrants ("second-generation Americans") differ from those of adult offspring of natives ("third-and-higher-generation Americans"). Policy and Research Context While New York traditionally has been a gateway for immigrants, the state faces new issues surrounding immigration and immigrant policy as a result of passage in 1996 of the Personal Responsibility and Work Opportunity Reconciliation Act (PRWORA) and the Illegal Immigration Reform and Immigrant Responsibility Act (IIRIRA). These new laws make the attainment of citizenship, rather than being a legal immigrant, a principal criterion for eligibility for public benefits. The welfare law, PRWORA, also shifts to the states new decisionmaking and fiscal responsibility for providing public benefits to immigrants. The new laws tighten restrictions on recent immigrants' eligibility for most major federal government benefits, severely limiting access to most public assistance during their early years in the United States. New immigrants are barred from Temporary Assistance for Needy Families (TANF) and Medicaid for their first five years and from Supplemental Security Income (SSI) and federal food stamps until they become citizens. These and other developments have heightened concern and public debate about the appropriate levels of immigration and immigrant eligibility for government programs. The findings from this study inform that debate in a number of ways, including: Fill Information Gap: To assess the potential implications of these changes requires information that until now has been missing. There are no official estimates of the total number of legal immigrants living in the United States, for instance, or of the number in each of the major subgroups of legal immigrants. Ironically, the only immigrant group for which official estimates exists is undocumented aliens (Passel 1997, Warren 1997). This study provides estimates of the size of these subpopulations.
Address Current Research and Policy Limitations: Fiscal impact studies and the policy debates surrounding their findings have been limited in at least four ways:
One-Sided Focus: Most research and public discussions have tended to focus on the costs of immigrants and their implications, with considerably less attention to the fiscal contributions of immigrants. For instance, in March of 1998, The Washington Times ran a series of articles focusing on the costs to county governments of providing services to immigrants and other negative impacts. None of the articles mentioned tax contributions of immigrants or immigrant-owned businesses. The focus of this report— the income and taxes paid by legal immigrants— helps to balance the research and policy debate.
Some recent research (Smith and Edmonston 1997) has used greatly improved methods to weigh the relative costs and revenues associated with immigrants. Yet even these efforts could be further improved by countering some of the inherent biases in the available data, which make it easier to measure use of services than tax payments. We address this issue by estimating total income and the following seven major taxes for individuals and households in New York State:
We also provide separate estimates for residents of New York City and the rest of the State. Lack of Critical Distinctions: Types of Immigrants. Most analyses of immigrants' fiscal impacts have failed to distinguish between undocumented and legal immigrants, and fewer still distinguish among types of legal immigrants. Even among legal immigrants, these distinctions can be critical because socioeconomic characteristics and eligibility for government benefits vary significantly by specific legal status. Refugees are eligible for more types of benefits than other aliens; green card holders face strict new eligibility restrictions under welfare reform; nonimmigrants remain ineligible for most government benefits; and undocumented aliens, already ineligible for most government benefits before welfare reform, face further restrictions. Despite these and other differences, analysts and policymakers often draw conclusions regarding the impacts of regularly admitted legal immigrants based on the characteristics of all noncitizens. These assessments underestimate the quality and contributions of these aliens since they have, on average, higher incomes than refugees or undocumented aliens.
A basic problem lies in using data sets that distinguish among natives, naturalized citizens, and noncitizens (aliens), but make no other legal status distinctions. We address this problem by developing a method that assigns exact legal status to all foreign-born individuals in the March 1995 Current Population Survey (CPS). The importance of this methodology is highlighted by the fact that two-thirds of New York's noncitizens in the March 1995 CPS are LPR aliens; one-third are not. Our methods allow, for the first time, detailed income and tax estimates for each of the five major immigrant status groups— naturalized citizens, LPR aliens, refugees and asylees, nonimmigrants, and undocumented aliens -- as well as natives. With our data and methods, analysts can use existing data sets to produce estimates of educational attainment, welfare use, and other characteristics essential for evaluating how welfare reform and other policies will affect the different immigrant populations. Distinguishing among legal status groups is crucial for assessing the fiscal and economic impacts of immigrants:
Levels of Government. Recent studies of fiscal impacts have been virtually unanimous in recognizing a fundamental asymmetry regarding the tax payments and service costs of both natives and immigrants. A large majority of tax payments flow to the federal government, while the most expensive services, especially public education, are paid for at the local and state levels. Our study finds exactly this pattern of tax payments in New York:
This imbalance is especially crucial under the new welfare reform regime. Welfare reform, particularly the reduction in food stamps for immigrants, shifts costs from the federal government to the state governments, because many states, including New York, have provided replacement food stamps to some immigrants who lost federal benefits. Nonetheless, immigrants continue to pay the bulk of their taxes to the federal government. In New York, distinctions between New York City and the rest of the state are also important because the characteristics of immigrants in each area differ. Outside New York City, the legally present foreign-born have higher incomes and pay more in taxes, on average, than natives. Inconsistent Attribution of Costs and Contributions: The failure to distinguish between second-generation Americans and other U.S.-born natives has led to serious inconsistencies in the research on fiscal impacts of aliens. Many assessments of immigrants' fiscal impacts use households rather than individuals as the unit of analysis (Smith and Edmonston 1997, Espenshade 1997), so the costs and contributions of immigrants' U.S.-born children are attributed to immigrants. Since children are expensive— having little or no income, paying little in taxes, but incurring large costs for public education— this treatment increases the estimated net costs of immigrants. However, in such analyses, when these U.S.-born children of immigrants become adults, establish their own households, and begin working and paying taxes, they are counted as natives. To be consistent, if second-generation Americans are counted as immigrants when they are young and expensive, then they should also be counted as immigrants when they are taxpaying adults. We provide the data needed to consistently assess the fiscal impacts of the children of immigrants by producing separate income and tax estimates for second-generation Americans and third-and-higher-generation Americans. Failure to Take the Long View: In general, the economic situation of immigrants appears to improve over time, whether measured over an individual immigrant's lifetime or over generations from the immigrants themselves to their U.S.-born children. In periods such as these, when a substantial portion of immigrants have arrived recently, failure to distinguish between short- and long-term immigrants paints a distorted picture of the probable long-term impacts of immigrants. Our study provides measures of adaptation by producing income and tax estimates for: (a) short- and long-term legal immigrants and (b) two groups of natives— the adult offspring of immigrants (second-generation Americans) and the descendants of U.S. natives (third-and-higher-generation Americans).(2) These measures allow assessments of the economic circumstances of recent immigrants, who under welfare reform face new restrictions on their eligibility for government programs. They also allow us to compare these circumstances with those of legal immigrants (again, by legal status) who have been here a longer period of time. From both perspectives, we find that immigrants in New York are adapting to the United States:
Data Sources Official data from the Immigration and Naturalization Service (INS), the Office of Refugee Resettlement (ORR), and the U.S. Bureau of the Census are used to estimate the total number of legal immigrants residing in New York and the size of each of the four legal immigrant categories. Estimates of the number of undocumented aliens are from the INS (Warren 1997). The principal data sources for the income and tax estimates are the March 1995 Current Population Survey as modified with the Urban Institute's TRIM2 computer simulation, the 1996 New York City Housing and Vacancy Survey, and a variety of administrative data sources. The separate population, income, and tax estimates that are given for New York City and the rest of the state arise from the survey data rather than independent demographic estimates. Overview of Results
Income
Taxes 
Policy and Research Implications Immigrants in New York, particularly legal immigrants, pay a significant proportion of taxes collected from New York residents. Our research points out some important distinctions that are often overlooked in discussions of immigrants and policies affecting them. Specifically, the differences among the legal status groups for immigrants are greater than differences between immigrants and natives. Further, the groups doing the worst economically (refugees and undocumented aliens) are generally not the groups affected by legal immigration policies being widely discussed. The results of our research and analysis prompt several observations. More Sophisticated Use of Data Needed: The findings underscore the need for a more sophisticated approach to the use of data for determining the fiscal impact of immigrants. At a minimum, separate assessments should be done for legal and illegal immigrants and, if possible, for each group of legal immigrants. When such distinctions are made, vast differences emerge in the various groups' economic situation, use of services, and, potentially, economic prospects. Analyses that fail to distinguish among the significantly different subgroups of immigrants miss these critical differences and may mislead policymakers who are developing new admission, settlement, and assistance policies. These concerns are particularly relevant where refugees and LPRs are concerned. Refugees. Among legal immigrants, refugees pay the least in taxes in absolute terms and as a percentage of their income. Unlike other immigrants, however, refugees are admitted for humanitarian reasons, without regard to their potential costs to the United States. Refugees often find it particularly difficult to work in the United States because of persecution in their home countries; stays in refugee camps; abrupt, unplanned departures from their homes; and any resulting physical and mental disabilities. Not surprisingly, as a group, they are substantially more reliant on government assistance than other immigrants, with about 20 percent of their income coming from welfare. Since the early 1980s, federal resettlement assistance for refugees has been cut from three years to eight months. Under welfare reform, refugees' eligibility for public benefits has been limited to their first five years for TANF and food stamps and first seven years for SSI and Medicaid. We find that refugees' incomes improve somewhat as their time in the United States increases. However, the improvement is relatively small and their incomes remain substantially lower than those of other legal immigrants (and even those of undocumented aliens). Many refugees, especially the elderly, may have difficulty finding other sources of support after their eligibility for government benefits expires, as mandated under the new laws. Legal Permanent Resident Admissions. Immigrants admitted as legal permanent residents are the focus of most debates about legal immigration policy because they represent the largest group of new entrants. However, most data sets used to describe immigrants do not distinguish immigrants admitted as legal permanent residents from other immigrants, although these other immigrants have substantially different characteristics. Analyses of fiscal and other impacts of legal immigration policy will be distorted unless researchers compensate for the presence of refugees and undocumented aliens, who have lower average annual incomes than immigrants admitted as legal permanent residents -- $8,300 for refugees and $12,100 for undocumented aliens -- versus $18,700 for immigrants admitted as legal permanent residents (which includes current LPR aliens and LPR aliens who have naturalized). (See Detailed Table 3).(3) Greater Analytic Consistency and Longer Time Perspective Needed: The findings in this report confirm that there is a need for greater analytical consistency in fiscal impact studies that compare natives to legal immigrants. In most analyses, the costs of native-born children of immigrants— principally public education and welfare costs— are attributed to their immigrant parents, but the tax payments of native-born adults whose parents were immigrants are attributed to natives. To be consistent, the fiscal impacts of second-generation Americans, whether they are children or adults, should be attributed to the same group— to immigrants or to natives. Counting second-generation Americans as immigrants when they are young and costly, but as natives when they start contributing taxes, biases analyses toward finding that immigrants are a fiscal burden. Our findings clearly show that as immigrants' time in the United States increases, their incomes and, consequently, their tax payments also increase. Recent welfare laws severely limit most immigrants' access to government benefits during their early years in the United States, when their incomes are lowest. While recent immigrants' access to most major federal government benefits was already restricted, the new welfare laws are more restrictive, barring new immigrants from TANF and Medicaid for their first five years and from food stamps and SSI until they become citizens. It is unknown whether the new bars on government assistance during immigrants' early years will push them to economic success faster or will slow their economic ascent. Federal-State Balance of Burdens and Responsibilities Needed: Finally, underlying much of the concern about the fiscal impact of legal immigrants is the historical relationship between the federal government and the states in sharing the costs and benefits of citizens. As noted earlier, the bulk of tax contributions from immigrants goes to the federal government, while large costs, especially those associated with public primary and secondary education, have fallen to the states and localities. Recent welfare laws have exacerbated the situation by shifting the burden of caring for new immigrants to the states. Until this imbalance is rectified, there will be continuing debate over whether to offer and how to pay for services used by legal immigrants. Ultimately, citizens will have to decide whether to raise state and local taxes, to petition for federal relief, or to curtail these services. Legal immigration policy and border control procedures raise even larger issues of intergovernmental equity. Constitutionally, the federal government has total authority to determine the total number of legal immigrants entering the country. Further, the characteristics of the entering immigrants are determined by federal policies on the mix of various categories of admission, including the level of humanitarian admissions. The control of undocumented immigration is also a uniquely federal responsibility, yet this group of immigrants poses special challenges to localities because of their low incomes, their impacts on school systems, and their concerns about interactions with government agencies and officials. Policy responses to immigration have largely failed to address these intergovernmental issues in any significant way. Yet, their fundamental nature underlies many of the ongoing debates about immigration and immigrant policy.
II. IntroductionThe Immigrant Population of New York Population Size. Throughout much of the country's history, New York State, especially New York City, has been the principal gateway for immigrants entering the United States. In recent years California has surpassed New York, although the State remains the second largest entryway to the country. At 3.4 million,(4) New York has the second largest immigrant population among the states, trailing only California, which has a foreign-born population that is more than twice as large, 8.1 million. (See Table 1.) New York has a significantly larger foreign-born population than Florida (2.2 million) or Texas (2.1 million), the states with the third and fourth largest immigrant populations. The foreign-born constitute 17.7 percent of the New York State population, the second highest concentration in the country, but well below the percentage in California, 25.1 percent. On this measure, New York ranks slightly higher than Hawaii (16.6 percent, not shown in Table 1), Florida (15.2 percent), and New Jersey (14.6 percent). With such a long-standing dominant role in the country's immigration, peaks and declines in New York's foreign-born population parallel trends for the nation as a whole. The foreign-born population of New York fell from a peak of almost 3.2 million in 1930 to just over 2.1 million in 1970 (Figure 1, page 20). With the new wave of post-1965 immigration, New York's foreign-born population climbed steadily to more than 3.2 million in 1996, reaching or even slightly surpassing the previous peak.(5) Since early in this century, the percentage of New York's population that is foreign-born has been almost exactly twice the national value. In 1996, 17.7 percent of New York's population consisted of immigrants, or slightly more than one person in six; nationally, about 9.3 percent of the population is foreign-born. A higher percentage of the New York population was foreign-born in 1996 than 1970, but the percentage remains well below the level in 1920, when 26.8 percent of New Yorkers were foreign-born (versus 13.2 percent nationally). Immigrant Origins and Legal Status. New York has a very diverse immigrant population, possibly the most diverse of any large immigrant state. Immigrants come from all parts of the world, and no region or country dominates. The Caribbean accounts for 915,000 of New York's 3.4 million immigrants (Table 2), or 27 percent; Europe has sent another 864,000 or 25 percent; and South and East Asia, 665,000 or 20 percent. The largest single country of birth for immigrants is the Dominican Republic, with 395,000 or 12 percent of all foreign-born. The next largest are: China(6) (229,000 or 7 percent), Jamaica (195,000 or 6 percent), and the countries of the former Soviet Union (182,000 or 5 percent). Among the other states with large immigrant populations, only New Jersey has a similarly diverse immigrant population (Espenshade 1997). In Illinois and Texas, the Mexican-born population predominates; Florida's immigrant flows are dominated by Latin American immigrants; and California has a dominant Mexican population, with extremely large secondary concentrations of Central Americans and Asians. Eighty-six percent of New York State's immigrants, 2.9 million, are legally present in the United States. (See Table 2.) Of legal residents, almost 1.2 million or 40 percent are naturalized citizens. This high percentage reflects the origins of New York's immigrants; large shares are from European countries or are long-term residents— both of which are associated with a high propensity to naturalize (INS 1997). Nonetheless, New York has the third largest number of undocumented aliens in the United States. The official INS estimate of 540,000 for New York places it third behind California (2,000,000) and Texas (700,000). (See Table 3(7) and Warren 1997.) About 3.0 percent of New York's population consists of undocumented immigrants (Table 1). This percentage is relatively high, as the only other states with more than 2 percent of their population undocumented are: California (6.2 percent), Texas (3.7 percent), Illinois (2.5 percent), and Florida (2.4 percent). The great diversity of origins among New York's immigrants is even more apparent in its undocumented population. The largest source country, the Dominican Republic at 36,000, accounts for less than 7 percent of the undocumented population; the country with the next largest share, Haiti, constitutes only 4.8 percent. The largest source region, the Caribbean, accounts for less than 30 percent of the total. In contrast, Mexican undocumented immigrants alone account for 69 percent of California's undocumented population, 77 percent of Texas', 65 percent of Illinois', and 90 percent of Arizona's. In Florida, 83 percent of the undocumented population is Latin American, versus 55 percent in New York. Distinguishing among Types of Immigrants Immigrants coming to live in the United States enter in one of three broad classes — legal permanent residents, refugees, and illegal immigrants.(8) These groups differ substantially from each other: they are admitted under different laws, are regulated by different agencies, have different socioeconomic characteristics, and qualify for different government benefits. As a result, their fiscal impacts differ as well. In addition, recent legislation draws new distinctions for program eligibility among immigrants on the basis of citizenship. Unfortunately, researchers, policy analysts, and politicians have consistently failed to distinguish among groups of immigrants on the basis of legal status. For example, undocumented immigrants are thought to have substantially less education and lower incomes than immigrants entering as legal permanent residents. Lower incomes translate into lower tax payments and higher public service costs. As a result, analyses that do not distinguish between legal and illegal immigrants will tend to overstate the decline in "quality" of recent immigrants and overstate the costs that legal immigrants impose on society. A major problem is that the data used in most analyses— decennial censuses and the CPS— do not distinguish immigrants in the United States with the nation's consent from those here without consent. The inclusion of undocumented aliens in these data sets is usually unrecognized by analysts. The 1980 Census included 14 million immigrants, about 2 million of whom were undocumented aliens (Warren and Passel 1987); likewise, the 1990 Census included about 20 million immigrants, over about 2.3 million of whom were undocumented immigrants (Passel and Kahn 1998). Our recent analyses suggest that, out of almost 25 million foreign-born individuals in the CPS in the mid-1990s, between 4 and 5 million were undocumented immigrants (Passel et al. 1997). It is therefore essential that analysts and policymakers recognize that the foreign-born population from the census and CPS cannot be used to represent characteristics of legal immigrants. For example, we estimate that 13 percent of immigrants in the CPS in New York are undocumented.(9) Unfortunately, even sophisticated analysts are not clear about this issue. The new National Research Council's report (Smith and Edmonston 1997) compares recent immigrants from Mexico (entering in the five years before the census) taken from the 1990 Census with those recent entrants enumerated in the 1970 Census. The report reiterates the oft-repeated finding that the data clearly showed a relative decline in the "quality" of immigrants (measured principally by education). From a policy perspective, the report, however, does not compare equivalent groups. While very few undocumented aliens appeared in the 1970 Census, roughly two-thirds of Mexican immigrants in the 1990 Census were either undocumented aliens or former undocumented aliens who had legalized under the Immigration Reform and Control Act of 1986 (IRCA).(10) The comparison tells us little about the consequences of current legal immigration policy, nor if it should be changed, because the analysis compares legal with illegal aliens. The major data sets used to assess the impacts of immigration, such as the decennial censuses and the CPS, include no questions on whether an individual is legally present in the United States. Thus, to account for legal status in using census or CPS data, analysts have found it necessary to resort to aggregate or proxy approximations to some legal status groups (for example, Fix and Passel 1994, Clark et al. 1994). Such methods generally treat characteristics of country-of-birth and period-of-entry categories as approximating the characteristics of legal status groups when, in reality, country-of-birth and period-of-entry groups include both legal and illegal aliens. The analysis presented here attempts to resolve this analytic dilemma by assigning legal status to individual CPS respondents. Legislation and regulations relating to immigrants' legal status and citizenship often seem to be based on the assumption that all persons in a family or household have the same legal status. However, many immigrant households contain people with different legal statuses. In fact, almost half (47 percent) of households headed by legally present foreign-born include one or more natives; even 37 percent of households headed by undocumented aliens include at least one native (see Table E, page 22). Put another way, about 30 percent of persons in households headed by legally present foreign-born are natives. These mixed households complicate policy options regarding eligibility for public benefits, particularly since many of the natives involved are children. Immigrant Costs versus Taxes In the last five years, there has been a resurgence of interest in studies of the fiscal impacts of immigrants. The bulk of analytic attention has focused on assessing the costs of providing services to immigrants, with less effort devoted to measuring the taxes paid by immigrants. (See Rothman and Espenshade 1992 for a review of earlier studies and Smith and Edmonston 1997 for more recent work.) When taxes were estimated, they were often based on extremely simplistic models (for example, Huddle 1993 and Internal Services Division 1992). More recent work such as Garvey and Espenshade (1997) and Smith and Edmonston (1997) has begun to use sophisticated methods to estimate both government program costs and tax receipts attributable to immigrants. Despite these advances, there are two major problems with this type of analysis. First, particularly in popular accounts, there seems to be an undue focus on the cost side of cost/benefit analyses of immigrants. For instance, in early March of 1998, The Washington Times ran a series of articles on the impacts of immigrants that focused on the costs to county governments of providing services to immigrants and other negative impacts. None of the articles mentioned tax contributions of immigrants or immigrant-owned businesses. Second, even sophisticated research often does not distinguish between legal immigrants, who are in the United States with knowledge and permission of the government, and undocumented aliens, who are in this country in violation of U.S. law. Furthermore, even among legal immigrants, research does not distinguish between regularly admitted legal immigrants and humanitarian admissions. Immigrants admitted for economic purposes and for family reunification are presumably expected to "pull their own weight" from a fiscal perspective, at least after a short adjustment period. Refugees and asylees, on the other hand, are admitted for humanitarian reasons, without regard to their potential costs to the United States. As noted above, the data sets used in these analyses, the decennial census and the Current Population Survey, contain large numbers of undocumented aliens. They also contain significant numbers of refugees and asylees. Since these groups have lower incomes, and therefore lower tax contributions, than regularly admitted legal aliens, using all noncitizens to represent the impact of regularly admitted legal aliens will seriously underestimate tax contributions for regularly admitted immigrants and will lead to misestimation of aggregate costs, and possibly average costs as well. The Importance of Generation: How Do We Count the Children of Immigrants? Researchers investigating the impacts of immigrants must deal with the U.S.-born children of immigrants. Although they themselves are natives, it can be persuasively argued that their presence in the United States, and therefore any associated costs (or benefits), should be attributed to immigrants. For analysis of fiscal impacts, there has been widespread agreement that the appropriate level of analysis is the household; with this approach, the costs of U.S.-born children of immigrants (that is, the second generation) who are still living at home are ascribed to their immigrant parents. Because public education is a major fiscal impact, perhaps the largest, the costs of these children, and therefore the households they live in, are high (Garvey and Espenshade 1997; also Clark et al. 1994). When these "expensive" native-born children grow up and leave their immigrant parents' households, they tend to become net contributors to the public sector. That is, they leave the public education system behind and enter the labor force. At this point and for a number of years until they leave the labor force, they pay taxes and use little in the way of services. Because of the methods used for estimation, adults of the second generation get counted as natives. Thus, the second generation is on the immigrant side of the ledger when they generate costs and the native side when they generate taxes. This basic imbalance biases fiscal impact studies toward finding net costs of immigrants. New work by Lee and Miller (1998) as reported in the National Research Council's assessment (Smith and Edmonston 1997) takes a different approach. Lee and Miller focus on the lifetime costs and taxes of immigrants. In a novel twist, they also attribute to immigrants the future costs and benefits of the immigrants' descendants.(11) With this approach, they find that the younger an adult immigrant is, the greater the net contribution, because younger immigrants have a longer working life to contribute tax payments. For immigrant children, lifetime contributions are positive, but the younger the child, the lower the net contribution because the public must pay the costs of education in the United States. Their other major finding is that the more highly educated the immigrant, the greater the net lifetime contribution. Lifetime contributions are positive for college-educated immigrants entering at ages younger than approximately 50 and negative for older immigrants. For immigrants with a high school education, the cross-over is about age 35; for less than high school, it is about age 20. With the current mix of immigrants, the average lifetime contribution of each immigrant and his/her descendants is about $80,000. It should be noted, in addition, that this figure is based on characteristics of immigrants in the CPS and thus includes legal immigrants, undocumented immigrants, and nonimmigrants. Consequently, it understates the value of each legal immigrant. Our report is concerned only with incomes and taxes, not with the costs of providing public services or net contributions. Nonetheless, because much of our analysis is household-based, in which we count incomes and tax contributions of native-born children with their immigrant parents, we also distinguish between two types of natives: second-generation Americans— those who have one or two immigrant parents; and third-and-higher-generation Americans— those with two U.S.-born parents. Overview of Methods Our research on immigrants in New York has three principal analytical tasks: (1) the development of population estimates to measure accurately the different immigrant populations and to weight the March 1995 CPS; (2) assignment of legal status to individual immigrants in the March 1995 CPS; and (3) estimation of the income and amount of each tax paid by individuals in the March 1995 CPS. The remainder of this section gives an overview of these methodological aspects of the work. Details of our methods can be found in Section IV, "Population Estimates" and Section V, "Tax Estimates." Population Estimates and Weighting. The CPS, which contains more than 4,000 households in the New York State sample, develops population numbers through a process of weighting individual respondents to be representative of each state's population. The weighting process is not designed, however, to give accurate estimates of immigrant populations. In this project, we remedy this deficiency by adjusting the CPS weights for immigrants in New York so that the weighted population figures more accurately reflect the number and characteristics of naturalized citizens, refugees, undocumented aliens, and legal permanent resident aliens. The weighting process relies on independent estimates of the number of immigrants in New York State in the various legal statuses, subdivided by countries and regions of birth. For all the populations except undocumented aliens, we develop population estimates for New York State using demographic techniques. For these groups (refugees, legal aliens, and naturalized citizens), we use data from the INS on the number of persons entering the country or the number naturalizing in each year. We then estimate mortality, out-migration, and migration between states for the period from the year of entry to 1995. The result is an estimate of the number of persons of each status residing in New York State. For undocumented immigrants, we use the official estimate of this group developed by the INS (Warren 1997). Immigrant Status. For this research, we assign legal status (naturalized citizen, refugee, nonimmigrant, undocumented immigrant, legal permanent resident alien) to each immigrant in the CPS. (See Table D for definitions.) The procedures differ for each status and vary greatly in complexity. For naturalized citizens, we use CPS responses to a direct question on citizenship in most cases. Some recently arrived immigrants report being naturalized citizens even though, according to the laws governing naturalization, they appear to be ineligible. We recode these individuals to be aliens unless they are married to a U.S. citizen or are the child of a U.S. citizen. We assign individuals as refugees if they are from refugee-sending countries and they arrived in a year when most of the arrivals from their country of birth came to the United States as refugees. For both nonimmigrant and undocumented status, our procedures are more complex and involve information on the immigrant's age, occupation, and relationship to other household members. For nonimmigrants, we match the characteristics of recently arrived immigrants in the CPS with the requirements for specific types of nonimmigrant visas. For example, J visas are given to scholars teaching or doing research on a temporary basis at U.S. universities. Thus, we assign a CPS respondent to be a nonimmigrant if the CPS respondent (a) had arrived in the United States within the last two years; (b) had an advanced degree; (c) was employed at a university; and (d) had a spouse who was neither a U.S. citizen nor employed full-time. To assign CPS respondents as undocumented immigrants, we employ a more complicated, probabilistic model. We first estimate the probability that a person of a given sex with a particular occupation is undocumented. This estimation uses a survey of formerly undocumented aliens at the time they acquired legal status under the Immigration Reform and Control Act of 1986 and survey data on the entire foreign-born population in 1995. For example, food service and household workers have a very high probability of being undocumented while judges cannot be undocumented. We then assign individuals to legal or undocumented status using these estimated probabilities. Finally, we employed a series of consistency checks and edits to screen out individuals who could not be undocumented. Aliens not assigned to any of the other statuses are assumed to be legal permanent resident aliens. Tax Estimates. The initial estimates of taxes paid and income for each household in the CPS were done with a microsimulation program developed by the Urban Institute called TRIM2 (for TRansfer Income Model Version 2). Essentially, this program fills out federal and state tax forms for a CPS household using the information collected in the CPS; the program estimates dependents, exemptions, and various deductions. TRIM2 also estimates the amount of Social Security tax (also referred to as the Federal Insurance Contribution Act tax or FICA) and unemployment insurance paid by household members and on their behalf by employers. TRIM2 tends to underestimate income and taxes for very high income households so that much of our work with the TRIM2 results involves adjustments for missed income at the upper end of the income distribution. We use the TRIM2 estimates of New York State income tax to estimate New York City income tax since, for most people, the two are directly related. To estimate residential property tax for New York City residents, we use the New York City Housing and Vacancy Survey, which collects detailed information on housing costs and characteristics, including nativity of the householder. With these results, we estimate property taxes paid by natives and immigrants and develop models to relate property tax payments to household characteristics. We then apply these models to allocate the property tax estimates to CPS households. Residential property tax estimates for residents of the rest of the state are based on New York State administrative data and CPS data. The final tax we estimate, state and local general sales tax, is estimated as a percentage of estimated household expenses on taxable items. We model household spending with a simple consumption model that makes allowances for housing costs, food costs, and immigrant remittances to their home countries. Finally, estimates of taxes paid by each household are allocated to individuals within the household based on their labor force status and personal income. III. Major FindingsDo the Legal Foreign-Born Pay Their "Fair Share" in Taxes? At the individual level, the legal foreign-born pay their fair share in taxes. (See Table B, page 7.) Average total tax contributions of the legal foreign-born overall ($6,300) are only slightly lower than those of natives ($6,500). Average tax contributions of naturalized citizens ($8,600) actually surpass those of natives, while average tax contributions of nonimmigrants ($6,400) are about equal to natives. Average individual taxes paid by LPR aliens ($5,000) and refugees ($2,200) are lower, especially for the latter group. (See Figure 2, page 28.) The average individual tax contributions of the legal foreign-born reflect their average individual incomes. Overall, their per capita incomes equal natives', about $18,000 for both groups. Per capita income is highest for naturalized citizens ($23,900) and nonimmigrants ($18,700), and is lower for LPR aliens ($14,500) and refugees ($8,300). Another approach to this question is to ask what percentage of income each group pays in taxes. Natives pay 30.7 percent of their income in taxes. For legally present immigrants, the percentage is only slightly lower at 29.1 percent (Table B). Refugees are much lower, 20.9 percent, and undocumented aliens lower still at 15.4 percent. The differences between legal status groups and natives are principally attributable to the progressive nature of income taxes, which result in higher income individuals paying a higher percentage of income in taxes. Since there are proportionately more natives at the highest income levels than legal immigrants, natives pay a slightly higher percentage in taxes, but both groups are paying roughly the same "share." At the household level, however, a somewhat different picture emerges. The average total tax contribution of households headed by the legal foreign-born ($13,300) is substantially lower than that of households headed by natives ($17,800). No legal foreign-born group, including naturalized citizens ($15,600) and nonimmigrants ($15,900), equals the average tax payments of native households (Figure 3, page 29). These somewhat lower household tax contributions reflect household income. Average household income for households headed by natives is $49,300, compared with $38,700 for the legal foreign-born overall. As is true for average household tax contributions, no legal foreign-born group, including naturalized citizens ($44,100) and nonimmigrants ($47,200), has average household incomes matching those of natives. In terms of percentage of income, natives and legal immigrants again pay roughly the same amount even at the household level. Native-headed households pay 30.7 percent of income in taxes; naturalized citizens, 30.8 percent; and LPR aliens, 28.4 percent. Refugee-headed households are again much lower at 20.9 percent because of the progressive effect of the income tax and the prevalence of nontaxable welfare income in these households. Why Do the Individual-Based and Household-Based Measures Differ? Differences between the individual- and household-based measures are largely due to the facts that a higher percentage of natives than legal foreign-born are children and that children have low incomes and tax contributions. Among natives, 29 percent are under 18 years of age. Among the legal foreign-born, only 10 percent fall in this age group. In the individual-level estimates, the relatively high proportion of children lowers the estimates for natives relative to the legal foreign-born. In the household-based estimates, the low income and tax contributions of U.S.-born children of immigrants are attributed to the foreign-born— their parents— rather than to natives. Altogether, 15 percent of households headed by the legal foreign-born and 13 percent of households headed by undocumented aliens contain native children. Finally, the incomes of adults in households headed by the foreign-born, especially undocumented aliens, are lower than the earnings for adults in households headed by natives ($18,400 for the legal foreign-born, $13,800 for undocumented aliens, and $26,000 for natives). The fact that adults in households headed by the foreign-born have relatively low incomes outweighs the fact that households headed by the legal foreign-born and undocumented aliens, on average, contain more adults aged 18 and over than households headed by natives (2.00 for the legal foreign-born, 2.12 for undocumented aliens, and 1.84 for natives). The difference in age composition between the native and immigrant populations can be dealt with by restricting the range of comparison of the populations or with age-standardization. The average incomes for individuals aged 18 and over show almost exactly the same patterns across immigrant status groups as do the household measures. Thus, adult immigrants have somewhat lower incomes than adult natives, but both legally present immigrants and natives pay roughly the same proportions of their incomes in taxes (Detailed Tables 1a-3a).
Who Lives in Immigrant-Headed Households?
Households headed by natives, for the most part, are composed entirely of natives; only 3 percent contain foreign-born individuals. The same is not true for the foreign-born. Among households headed by legally present immigrants, almost half (47 percent) include one or more natives; more than half (51 percent) of households headed by naturalized citizens have native members (Table E). Even in households headed by undocumented aliens, more than one-third (37 percent) have native residents. A large share of households headed by immigrants include native-born children. For both households headed by legally present immigrants and those headed by undocumented aliens, about 34 percent have native-born children. Of the legally present foreign-born, LPR aliens are the most likely to have native children in their households (42 percent), followed by naturalized citizens (30 percent). Few refugees have native children (5 percent) and, in our sample, no nonimmigrants do. Refugees and nonimmigrants are far less likely than the others to have U.S.-born children because they have, on average, been in the United States for fewer years. Although naturalized citizens have been in the United States for longer, on average, than LPR aliens, they are on average older, and therefore less likely to still have minor children in their homes. Households headed by the legal foreign-born are substantially more likely than households headed by undocumented aliens to contain U.S.-born adults. According to our estimates, about 20 percent of legal foreign-born households contain an adult native, compared with only 6 percent of undocumented alien households. Are the Incomes and Tax Contributions of the Legal Foreign-Born Outside New York City Similar to Those of the Legal Foreign-Born in New York City? Seventy-four percent of the legal foreign-born live in New York City, so the characteristics of the immigrants in New York City dominate statewide averages. In fact, outside New York City, where incomes of both natives and immigrants are higher than in the city, the average incomes and tax contributions of the legal foreign-born equal or surpass those of natives, no matter how they are measured. (See Table F.)
Unlike the state overall, outside New York City, the average total tax contributions of households headed by the legal foreign-born equal— and for some groups surpass— those of natives. For the legal foreign-born overall, average household tax contributions are $18,900, compared with $18,600 for natives. Average tax contributions for households headed by nonimmigrants ($22,300) and naturalized citizens ($19,200) surpass those of natives, while tax contributions of households headed by LPR aliens ($18,900) are about equal to those headed by natives. Tax contributions of households headed by refugees ($9,700) are lower. Tax payments of households reflect the general pattern of household incomes outside New York City, just as in the state as a whole. The share of income paid in taxes also follows the same patterns, but reflects the somewhat higher incomes outside New York City. Natives pay 30 percent of their income in taxes; legally present immigrants pay the same percentage. Naturalized citizens pay slightly more (31 percent) and refugees pay considerably less (22 percent). (See Detailed Table 3.)
How Do the Legal Foreign-Born Compare with Undocumented Aliens?
Although individual tax contributions of illegal aliens are not trivial— totaling over $1.1 billion— average tax contributions of undocumented aliens are substantially less than those of the legal foreign-born. (See Table G.) On average, undocumented aliens paid $2,400 in total taxes, less than half of the average individual tax contribution of the legal foreign-born. Household-based estimates are similar: the average household headed by an undocumented alien contributes $6,600 in total taxes, about half of the contribution of the legal foreign-born, $13,300.
The relatively low tax contribution of
undocumented aliens is due to two factors.
First, their incomes are substantially lower.
Per capita income for undocumented aliens is
$12,100, compared with $18,000 for the legal
foreign-born. Second, according to our
assumptions, tax compliance for
undocumented aliens is lower than for the
legal foreign-born. Although compliance is essentially mandatory for taxes such as property and
sales tax, we assume that compliance with federal, state, and local income tax, Social Security,
and unemployment insurance is 60 percent for those
undocumented aliens included in the CPS. Since even our
reweighted CPS does not fully represent the undocumented
population, this level of compliance is roughly equivalent to
50 percent compliance and leads to substantially lower
estimates of tax payments by undocumented aliens. (See
Section V, "Tax Estimates," below for explanation of
assumptions.)
How Does the Amount of Time Spent in the United States Affect Income and Tax Contributions of the Legal Foreign-Born? The longer immigrants have been in the United States, the higher their incomes and tax contributions. (See Table H and Figure 5, page 31.) We estimate income and tax contributions by period of entry for three groups— naturalized citizens, LPR aliens, and refugees— and for each, earlier entry groups have substantially higher incomes and, therefore, tax payments than later entry groups. For all periods of entry, the average individual incomes and tax contributions of naturalized citizens surpass those of natives. For LPR aliens, recent entrants (199095) have lower average individual incomes and tax contributions than natives, but long-term LPR aliens (entry before 1980) have average incomes that slightly surpass those of natives and equal tax contributions. The data in Table H support the notion that immigrants are integrating into American society. As a group, immigrants who have been in the country for 15 years or more have average incomes that exceed those of natives.(12) How Do Second-Generation Americans (Natives with One or Two Foreign-Born Parents) Compare with Third-and-Higher-Generation Americans? Incomes and tax contributions of second-generation Americans are remarkably similar to those of third-and-higher-generation Americans. (See Table I and Figure 6, page 32.) Among working-age adults (1864 years old), the average second-generation American has a total income of $26,800, compared with $26,900 for third-and-higher- generation Americans. Average individual tax payments are identical, $10,200. For older natives, those 65 and older, average individual incomes and taxes are somewhat higher for second-generation Americans than for third-and-higher-generation Americans: $20,100 versus $18,200 for incomes, and $4,900 versus $4,100 for taxes.
How Do Immigrants and Natives Compare on Different Taxes? The two basic methods of comparing natives and immigrants on taxes paid are (1) the proportion of taxes paid by the group relative to their proportion in the population; and (2) the proportion of a group's income that is paid in a specific tax. Both measures are displayed for every tax and each status in Table 10. Overall, natives represent 81.8 percent of the population and receive 82.6 percent of personal income. The $98.2 billion that natives pay in taxes represents 83.6 percent of the taxes paid— a higher percentage of taxes than their representation in the population. Natives pay a proportionately higher share of tax for federal income tax, state income tax, residential property tax, and sales tax. For Social Security tax, the proportion paid by natives, 81.9 percent, is only 0.1 percent higher than their percentage in the population and lower than their share of income. For unemployment insurance, natives pay a lower percentage, 80.6 percent. The proportion of tax paid by natives is largely a function of the progressivity of the tax. Natives have a disproportionately high percentage of very high incomes. The tax rate on income is higher for high incomes than low incomes— that is, the income tax is progressive, with federal income tax being more progressive than New York State's, which in turn is more progressive than New York City's. Thus, natives pay a higher percentage of the income taxes than of the other taxes. On the other hand, Social Security tax and unemployment insurance do not tax income above a certain amount— that is, these taxes are regressive. On these two taxes, natives pay a lower percentage than on the other taxes. Overall, natives pay a slightly higher percentage of their income in taxes, 30.7 percent, than do legally present immigrants, 29.4 percent; naturalized citizens pay 30.8 percent, a slightly higher percentage than natives, however. Again, the progressive nature of the different taxes is reflected in this measure. Legally present immigrants pay a higher percentage of their income than natives in the more regressive taxes, Social Security and unemployment insurance; on the more progressive income taxes, natives pay a higher percentage. Naturalized citizens pay a higher proportion of their income in taxes than any of the groups, reflecting their relatively high incomes and very high residential property tax payments.
Legal immigrants pay $12.5 billion in taxes to the federal government, or 68.9 percent of the total taxes they pay (Table J). Natives pay roughly the same percentage, 70.5 percent. The lowest percentage is for naturalized citizens (66.7 percent) because a disproportionate share of the taxes they pay goes to residential property tax. LPR aliens pay a very slightly higher percentage, 71.5 percent, of their taxes to the federal government than do natives. This higher percentage is somewhat ironic in that the federal government has limited access to public benefits for legal aliens, whereas many states, including New York, are filling some of the gaps left by the federal limitations. Refugees and nonimmigrants also pay a slightly higher percentage to the federal government than do natives. The percentage of taxes paid to the federal government is related to the progressive nature of federal taxes compared with state and local taxes. Two federal taxes are regressive (Social Security and unemployment insurance), while one (federal income tax) is quite progressive. In general, natives pay proportionately more of progressive taxes while immigrants pay more of the regressive taxes. The regressive and progressive taxes roughly balance each other out, so, across the status groups, the percentage of taxes paid to the federal government differs little. IV. Population EstimatesDeveloping population estimates for immigrant groups in New York State involves three interrelated and overlapping procedures. The first step is to develop estimates of the number of persons in each legal status group in New York State. The second is to assign immigration status to foreign-born individuals in the March 1995 CPS data set. And the third is to assign weights to the CPS individuals so that the estimates developed from the CPS agree with the independent population estimates. For this project, we estimate the size of the following groups of immigrants living in New York in 1995:
Entered 198089
Entered 198089 Entered before 1980
Entered 198089 Entered before 1980
The sample size and coverage of the CPS limit our ability to produce statistically defensible and meaningful estimates of income and taxes paid for all of the above groups in New York State. Where the CPS includes insufficient numbers of cases for a group (usually a weighted estimate of 50,000 persons or 25,000 households), we collapsed the period-of-entry groups into larger aggregations. For example, because the number of naturalized citizens who entered the United States between 1990 and 1995 is so small, this group is combined with the 1980-1989 entry cohort in all tables. Estimation of Population Size Estimates for the various immigrant populations (legal permanent resident aliens, naturalized citizens, refugees, and undocumented aliens) are produced for 35 separate countries or areas of birth (Table K) by age, sex, and period of entry. The estimates use the basic logic of demographic estimation; that is, for each immigrant entry cohort, the size of the group diminishes each year as group members die or leave the United States. In addition, we produce the estimates for the six states with the largest immigrant populations (California, New York, Texas, Florida, Illinois, and New Jersey) and the balance of the United States. For individual states, the size of each immigrant cohort is also affected by interstate migration. Components of Change. The same basic rates of change for mortality, emigration, and internal migration (modified where appropriate) are applied to each individual immigrant status group. For emigration and mortality, the rates are specific for race, sex, and age. For internal migration, sample sizes in the census were too small to estimate separate rates by age and sex, so the same rates are applied to entire country of origin groups. Mortality. The incoming immigrants and refugees, initial populations, and persons naturalizing are survived forward (by age and sex) to 1995 using as the base year the immigrants' year of admission (not the year of arrival) or year of naturalization, points at which the immigrants are known to be alive. The survival rates are calculated from the life tables (by sex and race/ethnicity) used in the Census Bureau's 1995 population projections(13) for race/ethnic populations (white, black, Asian/Pacific Islander, American Indian, Hispanic). For each country of origin, we used the race and ethnicity of the majority of immigrants to select a life table. Emigration. Measuring emigration among immigrants remains a difficult issue. The basic set of rates used in this study comes from Ahmed and Robinson (1994), who calculated sex and country-of-birth specific rates using 1980 and 1990 census data. The rates estimated by Ahmed and Robinson are for a ten-year period for five-year age groups. Our estimation process requires both longer and shorter exposure periods, different age groups for each year, and sex-specific rates. Iterative fitting is used to provide rates by single years of age for single years of time that are consistent with Ahmed and Robinson's ten-year rates for five-year age groups. For example, applying our single-year emigration rates to a cohort five times results in the same number of people leaving the United States as applying their five-year rate once. Ahmed and Robinson calculate both full and half exposure rates for the 1980s. Consistent with their estimates and observations, we use lower emigration rates for immigrants from Hispanic countries. The lower emigration rates are also used for naturalized citizens and refugees, groups with greater attachment to the United States and weaker ties to their home countries. The smoothed emigration rates (by age and sex) are applied to the surviving immigrant populations. Interstate migration. Country-specific rates of interstate migration for a five-year period are derived from 1990 Census data on residence in 1985 for immigrants who entered the United States before 1985. These rates provide a full matrix of flows to and from each of the seven areas. These five-year rates are applied three times to 1980 populations, twice to 198084 immigrants, and once to 198589 immigrants for the surviving and nonemigrating immigrants. Refugee Population. Refugees are admitted to the United States for humanitarian reasons; for example, they are fleeing persecution because of their religion, race, ethnicity, or political beliefs. Asylees are persons already in the United States who are granted legal status for the same types of humanitarian reasons. Refugees and asylees generally do not remain in refugee/asylee status for more than one to two years. They generally "adjust their status" and become legal permanent residents. Refugees and asylees differ from most other legal admissions because they are not admitted for their skills or because of their relationship to someone already in the United States. Since they are often "in flight" and ill-prepared and unequipped for life in the United States, they are the only large groups of legally admitted aliens eligible for welfare and other social services upon admission to the United States. They generally retain their eligibility for public benefits for several years even if they become legal permanent residents or naturalized citizens. For this study, we define refugees as persons admitted as refugees and asylees after 1980, regardless of their current status.(14) The detailed initial population is derived principally from tabulations of INS public-use microdata files on legal permanent residents. From these files of individual records for fiscal years 198095, we tabulate persons (and their dependents) who were adjusting from refugee or asylee status to legal permanent resident status; the tabulations are done by age, sex, country of birth, and state of residence. The INS files include no information on refugees and asylees who do not change their status to legal permanent resident. Almost all refugees and asylees eventually adjust their immigration status to legal permanent resident; most do so within one or two years of entry. However, some never adjust, and recent entry cohorts have not had a chance to fully adjust. Therefore, to take into account refugees and asylees who had not yet adjusted status, we compare our tabulations of the number of persons adjusting status by year of entry and country of birth with the total number of refugee and asylee arrivals derived from INS Statistical Yearbooks (various years) and annual reports from the Office of Refugee Resettlement (various years). The differences, representing unadjusted refugees and asylees, are added to the total counts of refugees and distributed by age and sex. In the final step of the estimation process, the refugees are survived to 1995 with the survival rates described above, and emigration and internal migration rates are applied. For 1995,(15) we estimate that there were 1,845,000 persons living in the United States who entered the country since 1980 as refugees or asylees (Table 4). New York contains 198,000 refugees, 10 percent of the national total. Unlike the United States as a whole, New York's refugee population is dominated by refugees from the former Soviet Union, who account for almost two-thirds (125,000) of the state's refugee population. Nationally, no country of origin group constitutes such a large share of the refugee population, and refugees from the former Soviet Union make up only 20 percent of the total. The largest source of refugees for the nation as a whole is Southeast Asia, 46 percent of the total, most of whom are from Vietnam, 28 percent of the total. In New York, Southeast Asians constitute a relatively small share of the refugee population, 11 percent, although, congruent with the nation as a whole, most are from Vietnam. Legal Permanent Residents. Our estimate of the legally resident foreign-born population (other than refugees) in 1995 is built from three major immigrant flows: (1) legal permanent residents entering in fiscal 1980 through 1995; (2) persons acquiring legal status through two legalization programs of the Immigration Reform and Control Act of 1986 (IRCA)— for pre-1982 entrants and for farmworkers; and (3) legal foreign-born residents in 1980. Each of these populations, by age, sex, country of birth, and state of residence, is subjected to mortality rates, survival rates, and internal migration rates as described above. The resulting surviving populations are summed to give the total number of legal foreign-born residents (other than refugee entrants and nonimmigrants) in 1995; we refer to this population as legal permanent resident (or LPR) admissions to distinguish it from legally admitted nonimmigrants, the refugee/asylee entrants, and undocumented aliens. The LPR admissions are further subdivided into naturalized citizens and legal permanent resident aliens using methods described below. The numbers of legal permanent residents entering in fiscal 1980 through 1995 are obtained by tabulating public-use microdata files from the Immigration and Naturalization Service. Immigrants adjusting status from refugee and asylee status are excluded from this tabulation because they are included in the refugee tabulations (described above). The tabulations are done by age, sex, country of birth, and intended state of residence. Immigrants with unknown state of residence are distributed according to the distribution of known state of residence for each country of birth. For fiscal years 1980 and 1981, the data on intended state of residence are missing for most individuals. For these years, for each country of birth, the immigrants are distributed using the average distribution of surrounding years. The estimation procedures for aliens legalizing under IRCA are similar to the procedures used to estimate legal permanent residents. The initial populations come from tabulations of INS public-use microdata files for persons legalizing as Special Agricultural Workers (SAWs) and those legalizing as pre-1982 entrants (known as LAWs). Both INS files provide information on date of most recent entry to the United States, which we treat as the date of entry for tabulation purposes, and information on age, sex, country of birth, and state of residence. One issue we had to resolve was the number of legalizing aliens who actually were granted permanent resident alien status. The data files had information on whether the applications for legal status had been approved, denied, or were pending. In addition, the INS Statistical Yearbooks for 1989 through 1996 provide information on the number of LAW and SAW applicants who eventually became legal permanent residents. To get a detailed distribution of the legalized population that ultimately acquired legal permanent resident status by age, sex, period of entry, country of birth, and state of residence, we produced two tabulations from each of the LAW and SAW files. One tabulation includes only those persons whose applications had been approved by the date of the files (November 1991), and the other includes this group as well as persons with applications still pending. We then interpolated between these two tables using the INS information on the total number eventually approved and acquiring LPR status.(16) For applying survival, emigration, and internal migration rates, all legalizing aliens are assumed to have been in the country in 1988 (the average date of application for legalization). Thus, these groups are exposed to seven years of mortality and emigration risk and one period of internal migration risk. Data on the legally resident foreign-born population in 1980 are drawn from Warren and Passel (1987) and Passel and Woodrow (1984). Unpublished tabulations consistent with the published material are used to provide the required information by age, sex, period of entry, country of birth, and state of residence. For some countries, detailed information from the 1980 estimates is not available; for these countries, estimates for larger, more inclusive areas are subdivided using 1980 and 1990 Census data.(17) Nationally, the legal permanent resident population— the sum of surviving post-1980 LPR admissions, 1980 legal residents, LAWs, and SAWs— is estimated at 17.5 million for 1995. The estimated number of survivors of the pre-1980 legal entrants, at 7.51 million, is very slightly larger than the estimated number of post-1980 legal permanent residents, 7.47 million. The surviving LAW and SAW populations are much smaller, at 1.55 million and 1.01 million, respectively (Table 4). The estimates for New York State reflect the state's history as a major destination for immigrants in the past (so that the 1980 legally resident population was older than the U.S. average), as a destination for recent flows, and as not having an overly large number of SAW legalizations. Thus, the estimated 1995 legal permanent resident population of 2.60 million has more post-1980 LPRs (1.40 million) than pre-1980 legal entrants (1.07 million). LAW residents (94,000) outnumber SAWs (32,000) by almost three-to-one. (See Table 4.) Naturalized Citizens. Constructing estimates for the naturalized citizen population follows procedures similar to those used for the legal permanent resident population. There are three major "flows" of naturalized citizens: (1) persons naturalized by the INS for fiscal years 1980 through 1995; (2) the resident naturalized citizen population as of 1980; and (3) persons deriving citizenship by naturalization of their parents or through marriage to a U.S. citizen for fiscal years 1980 through 1995. Each of these populations, by age, sex, country of birth, and state of residence, is subjected to mortality rates, survival rates, and internal migration rates as described above. The resulting surviving populations are summed to give the total number of naturalized citizens in 1995. (See Passel and Clark 1997 for more details of the estimation procedure.) The numbers of persons naturalized in fiscal years 1980 through 1992 are obtained by tabulating public-use microdata files from the INS by age, sex, period of immigration, and country of birth. For fiscal years 19931995, totals by country of birth and year of immigration are derived from published and unpublished INS tabulations; the totals are distributed by age and sex using the distributions from 19901992 by country of birth and year of immigration. For each year, the totals are distributed across states using INS data on naturalizations by state and country of birth. Estimates of the naturalized citizen population in 1980 by state of residence, age, sex, country of birth, and period of entry are from Warren and Passel (1987) and Passel and Woodrow (1984). Unpublished tabulations are used because they provide more detail than the published tables. Finally, because detailed information is unavailable for some countries, tabulations from the 1980 census are used to subdivide estimates for larger areas. Accounting for derivative citizenship is the most difficult part of the estimation process. Almost all persons receiving derivative citizenship certification from the INS became citizens upon the naturalization of their parents, but did not receive official certification of naturalization at the time. They do not appear in INS records until they get this certification and, even then, are not included in INS counts of naturalizations. For most years in the 1980s, the INS Statistical Yearbooks provide information on the number of derivative citizenships, their countries of birth, and the year the citizenship was derived. For these years, we assumed that the distribution by age, sex, and period of entry for derivative citizens from each country is the same as the distribution for children naturalized from that country in the year their citizenship was derived. In some years in the 1980s and most in the 1990s, the INS did not publish detailed information on derivative citizenship. For these years, we obtained unpublished information from the INS on the number of derivative citizenships granted. We allocate the totals to countries of birth and year their citizenship was derived by extrapolating from the years when detailed data were available; the derivation of detailed age-sex-period of entry estimates then follows the same procedures as before. For each year, the estimates are distributed to states using the distribution of naturalizations in that year for each country. Each of the three subpopulations of naturalized citizens are survived to 1995 and subjected to risks of emigration and internal migration (as described above for the foreign-born population). Risks of emigration are smaller than those for the entire foreign-born population because of the greater attachment to the United States shown by the choice of naturalizing. The three estimates are summed within age, sex, period of entry, country of birth, and state of residence groups to yield the final estimates. Nationally, we estimate there are 7.3 million naturalized citizens (Table 4), with the largest numbers from the Philippines (590,000), Mexico (486,000), China(18) (486,000), Cuba (375,000), and Vietnam (342,000). New York had the second largest state population of naturalized citizens at 1.2 million; California with 1.8 million had the largest number. The principal countries of birth for naturalized citizens in New York are China (95,000), Dominican Republic (80,000), countries of the former Soviet Union (59,000),(19) and Jamaica (56,000). (See Table 2.) Undocumented Aliens. For our estimate of the number of undocumented aliens living in New York, we draw on the latest official estimates of the Immigration and Naturalization Service (Warren 1997). The INS provides national estimates by country of birth for October 1996 and state level estimates for the total number of undocumented aliens; in addition, the INS estimates provide measures of average annual change over the period October 1992October 1996. Unlike previous INS estimates of the undocumented population (Warren 1994), the 1997 INS estimates provide no state-specific information on the countries of origin of undocumented aliens. We use the 1994 INS estimates as a basis for estimating state-specific estimates of the countries of origin of undocumented aliens for later periods. The INS estimates that 5.0 million undocumented aliens lived in the United States as of October 1996. New York had the third largest number at 540,000 or 10.8 percent of the total, behind only California (2.0 million or 40 percent) and Texas (700,000 or 14.1 percent) and ahead of Florida (350,000 or 7.0 percent) and Illinois (290,000 or 5.8 percent). (See Table 3.) The average annual increase was estimated at 275,000 for the United States and 33,000 for New York State. Using these figures to interpolate, we estimate the undocumented population of New York at 507,000 as of October 1995 and 488,000 in March 1995. Using iterative proportional fitting, we develop estimates of undocumented aliens by state and country of birth consistent with the INS estimates. The national estimates by country of birth and the state estimates of total undocumented aliens, both from the 1997 INS estimates, are used as marginal totals for a two-way table (Warren 1997). The initial estimates of the interior cells of the table (that is, state of residence and country of birth cells) come from the INS' 1994 estimates (Warren 1994). The distributions are alternately adjusted to national country-of-birth totals across states (the seven largest undocumented populations and the balance of the country) and then to state totals across countries. The process converged after six iterations. The estimates of undocumented aliens by state for October 1995 are shown in Table 3. New York State differs from the other major undocumented alien states in that no single country of birth dominates as a source of undocumented aliens. The largest estimate for any single country is Jamaica, estimated to have 33,000 undocumented aliens in New York, or 6.5 percent of the state's total. In other states, the largest single source represents 3 to 14 times this percentage. Assignment of Legal Status The foreign-born population— identified in the CPS through a series of questions on country of birth and citizenship status— is defined as persons born outside the United States who are naturalized citizens or not U.S. citizens (that is, aliens) and who were not born to U.S. citizens. The only information collected on legal status is whether an individual has become a naturalized citizen, and even this information is misreported by a significant portion of respondents (Passel and Clark 1997). However, our research has shown that it is possible to assign legal status with a series of editing and imputation procedures so that the numbers and characteristics of the CPS populations agree with administrative data on the size of the various populations. This section sets forth in brief the procedures used to assign legal status to the 1995 CPS cases for New York State.
This method produces a population that substantially overlaps with the actual refugee population. Further, it is likely that persons erroneously included as refugees share many of the socioeconomic characteristics of refugees, so estimates of refugee population characteristics should be accurate. For the 198095 period, 1.8 million persons were admitted as refugees or asylees. The total number of admissions of refugees, asylees, and legal permanent residents for the 17 "refugee countries"(23) shown in Table L for the "refugee periods" amounts to 1.9 million persons (Table 4), a figure just 9 percent larger than refugee and asylee admissions. Thus, the procedure gives a good approximation to the total number and sources of refugees. On the other hand, 1.6 million of the 1.9 million persons identified by the procedure, or 81 percent, were actually admitted as refugees or asylees. We apply the assignment procedure to the March 1995 CPS for the nation to estimate an initial refugee population of 1.9 million persons nationwide from the countries shown in Table L. This figure is within 1 percent of the estimated total number of refugees and asylees living in the United States in 1995. For New York State, the procedure assigns 151 cases (unweighted) in the March 1995 CPS as refugees. Using their initial weights (unadjusted), the estimate of refugees in New York State is 246,000. This estimate is reduced to 198,000 persons by adjusting the refugees' weights (Table 2). Naturalized Citizens. Passel and Clark (1997) compared INS records on naturalization with CPS data on naturalization and found that the number of "naturalized citizens" in the CPS is too high, apparently because substantial numbers of noncitizens falsely report that they have naturalized. False reporting of citizenship also appears to be a problem in the census. This overreporting of citizenship is attributable to two groups. Among persons who entered in the last five years, approximately three-quarters of those who report being naturalized citizens are not. Among long-term immigrants from Mexico and Central America, approximately one-third of those who report being naturalized citizens are not. Reported citizenship by other groups in the CPS — that is, long-term immigrants from other countries — appears to be accurate. We address the first part of this reporting problem by editing reports of naturalized citizenship so they conform with federal law. Persons who had lived in the United States for five or fewer years and reported they are naturalized citizens were recoded to be noncitizens (aliens) unless they met one of two conditions: (1) they had lived in the United States for at least three years and were married to a U.S. citizen; or (2) they were the child of a U.S. citizen. Nationally, this procedure reduces the number of naturalized citizens who entered after 1990 from 329,000 to 103,000, extremely close to our estimated total number of 91,000 (Passel and Clark 1997). For New York State, the editing procedure affects very few cases, reducing the CPS figure for naturalized citizens arriving after 1990 from more than 30,000 to 14,000, almost exactly the estimated actual total. No further adjustments to reported citizenship were made because, in New York, citizenship appears to have been accurately reported among immigrants who entered before 1990. In fact, the number of naturalized citizens in the New York CPS who entered before 1990 is slightly less then our estimate of the actual number of naturalized citizens for this period. Nonimmigrants. Nonimmigrants are aliens admitted to the United States for limited time periods and for specific purposes. More than 20 million nonimmigrants are admitted to the United States every year; by far the largest number are tourists. Some nonimmigrants are permitted to work in the United States, while others are not. For the purpose of a survey such as the CPS and for population estimates, the key feature of nonimmigrants is whether they can be considered U.S. residents. For the estimates presented here, we consider a nonimmigrant to be a U.S. resident if he or she appears in the CPS and meets one of the criteria described below. It should be noted, however, that nonimmigrants who earn income in the United States are generally required to pay tax on the income regardless of their residency status. In addition, any nonimmigrant who spends money in a state with a sales tax will pay that tax. The procedures used in this project match the characteristics of aliens in the CPS with the criteria used to award nonimmigrant visas.(24) The first criterion is that the nonimmigrant must have come to the United States within the last three years.(25) Next, if an alien's spouse is a U.S. citizen (either native or foreign-born) or the alien's parent or grandparent is in the household, the alien is not assigned to be a nonimmigrant. Finally, aliens who are self-employed or who work for the federal, state, or local governments are not assigned to be nonimmigrants. Aliens who are spouses or children of nonimmigrants and who entered the United States at the same time as the nonimmigrant or more recently are assigned as nonimmigrant dependents. According to the editing and assignment procedures, there were 696,000 nonimmigrants in the March 1995 CPS (Table 4); of these, 58,000 were in New York State (Table 2). The specific types of nonimmigrants and the editing rules follow (with the national total, including dependents, from the March 1995 CPS in parentheses):
2. Visiting Scholars, J visas ( 76,000) — teacher, social scientist, or scientist; employed at a university or research organization; spouse not working full-time. 3. Physicians, JL visas (38,000) — physician, health diagnosing professional, or therapist; not in a private or group practice; spouse not working full-time. 4. Nurses, H1A visas (18,000) — employed in a hospital or university as a nurse; spouse not working full-time. 5. Religious workers, R visas (3,000)— priest or other religious worker. 6. Intracompany transfers, L visas (140,000) — high income ($35,000 or more); over age 30; occupation limited to certain occupations (for example, administrator, manager, computer analyst), with certain other occupations not permitted (for example, postmaster, hygienist, real estate agent, sales). 7. Diplomats, A1 visas (34,000) — employed by foreign government; spouse not working. 8. High school exchange students, J visas (10,000) — nonrelative of household head; aged 1419 years; enrolled in high school; and there is a child of the household head of the same sex in the household. 9. Au pairs, H2B visas (5,000) — nonrelative of household head; child care worker; and there is a child or children of the household head under age 12 in the household. These procedures are designed to identify the aliens most likely to be nonimmigrants, not to identify every nonimmigrant in the CPS. Further, since the CPS is limited to the civilian noninstitutional population, not all nonimmigrants are included; for example, foreign students living in college dormitories are not covered. Because nonimmigrants may not consider themselves to be U.S. residents, even if they meet the technical definition, many are likely to be omitted from the CPS. Thus, these estimates should be considered as lower bounds on the number of nonimmigrants in the country. Undocumented Aliens. We developed a new procedure that assigns the probability of being an undocumented alien to each noncitizen in the CPS. Our choice of data was limited because very few data sets identify whether immigrants are undocumented aliens. However, in conjunction with IRCA's two legalization programs, which granted legal status to about 2.7 million formerly undocumented aliens, the INS conducted the Legalized Population Survey (LPS) to college collect information on applicants for the larger of the two legalization programs. IRCA allowed two large groups of undocumented aliens to legalize, approximately 1.7 million individuals who had resided in the United States for five or more years — that is, LAWs, who were admitted under Section 245A and were the focus of the LPS — and approximately 1 million agricultural workers. We used LPS data on the occupational structure of Section 245A applicants who lived in New York State at the time they legalized (when they were still technically illegal) to assign legal status to aliens in the 1995 March CPS. The procedure assigns a probability of being undocumented to individuals in the CPS on the basis of their age, sex, and occupation. The major underlying assumption for this procedure is that the occupational structure of undocumented aliens is circumscribed by their legal status and, therefore, differs significantly from that of legally resident aliens. Other major assumptions involved in the procedure are: 1. Aliens who legalized under Section 245A of IRCA are representative of all illegal aliens. 2. Legal and illegal aliens in the same state, with the same labor force status and occupations, in the same age group, and of the same sex have roughly the same socioeconomic characteristics. 3. Between 198788— the years in which IRCA aliens applied to become legal U.S. residents— and the mid-1990s, the occupational structure of illegal aliens did not change significantly. 4. The coverage of illegal aliens from a given country of origin in the CPS does not differ substantially by occupational status. However, as described below, we did allow for some variation by occupation group in the percentage of illegal aliens represented in the CPS. In fact, aliens who legalized under Section 245A of IRCA are probably somewhat better off than undocumented aliens overall because, at the time of legalization, they had been in the United States for at least five years. If our first two assumptions are incorrect, we will tend to overestimate the economic characteristics of undocumented aliens. It is not clear how violations of our last two assumptions might bias results. Estimation procedure: Occupation-based estimates. We estimate the number of illegal aliens in each major occupation group in New York State in March 1995 using the INS estimate of the number of illegal aliens in the state (Warren 1997) and the occupational distribution from the LPS. The occupation groups are the CPS' 14-category major occupation groups with certain occupations broken out separately (Table M). We use detailed occupation groups for occupations in which illegal aliens either are not represented (for example, health diagnosing occupations such as physicians and dentists, lawyers and judges, and public protective service occupations such as police, fire fighters, and crossing guards) or are substantially overrepresented (for example, retail sales workers and food service workers). Individuals who did not work in the past year are placed in a separate "not employed" category. Calculations are done separately for each sex.  We estimate the number of illegal aliens in each major occupation group as follows:
 where
 = estimated number of illegal aliens, i, aged 1864 of sex s in New York in occupation x in March 1995;
The next step explicitly adjusts for the fact that not all illegal aliens are captured by the CPS. We calculate the percentage of illegal aliens who appear in the March 1995 CPS by comparing the INS estimate of the total undocumented population with our estimate of the number of undocumented aliens included in the CPS. To derive the latter figure, we subtract the estimated number of legal foreign-born residents (derived with data on legal entrants and components of population change, as described above) from the CPS foreign-born population. According to our calculations, 82 percent of illegal aliens in New York in March 1995 are represented in the CPS. For each major occupation group, we calculate the percentage illegal among the aliens who appear in the 1995 CPS, as follows:
 = total number of noncitizens, T, of occupation x aged 1864, of sex s who
appear in the March 1995 CPS.
Not all aliens are considered to be potential illegal aliens. Aliens were also considered to
be legal, and therefore are not used to calculate
2. They had been assigned to be refugees or nonimmigrants in earlier estimations; 3. They reported working for local, state, or federal government; 4. They reported being veterans; 5. They reported receiving Supplemental Security Income (SSI) or being covered by Medicaid; and 6. They were in a household receiving food stamps (only reported at the household level) and in which all individuals are aliens. For some occupation groups, our initial estimate of the number of illegal aliens we expected to find in the CPS was greater than the total number of aliens in the occupation group. We proportionately adjust the number of illegal aliens expected to be found in the other occupation groups so the total representation of illegal aliens in the CPS is approximately 82 percent. The estimated probability that an alien is undocumented is based on his or her sex and occupation group. Within each occupation and sex group, the estimated probability of being undocumented equals the estimated proportion of aliens who are undocumented aliens. Before making the probabilistic assignments, we create duplicate records for each household so that we could produce refined estimates and carry out the consistency checks described below. The number of duplicate household records equals the household weight divided by 15, resulting in an average of roughly 100 duplicate records for each household with at least one potential undocumented alien. Individual, family, and household weights are adjusted so that, for each individual, the sum of the weights of the duplicated records is equal to the original weight.(26) For each alien in the new database, we generate a random number between 0 and 1. If this number is less than or equal to the estimated probability that someone with his or her occupation is an illegal alien, the individual is assigned to be an illegal alien; otherwise, the individual is assigned to be a legal alien. Estimation procedure: Household and family consistency adjustments and assignments. After all individuals are assigned an initial legal status, we check to make sure that individuals within a household have consistent legal statuses. The consistency checks and corrections are as follows, in the order they are done.
2. All alien parents and siblings of a minor alien child receiving Medicaid or SSI are assigned as legal. Since these children are legal and children cannot legally immigrate without their parents, the parents must be legal. Then, since we know the parents are legal, we assume all their other children are as well. 3. All alien household members who immigrated with or before an immigrant we know to be legal (because that alien is a government worker, veteran, receives SSI or Medicaid, or is a naturalized citizen, refugee, or legal nonimmigrant) are assigned as legal aliens. We decided it is implausible that household members who entered at the same time did not have the same legal status. We also decided it was implausible that long-term undocumented aliens would end up sharing households with more recent legal aliens, although we did allow for long-term legal aliens to share households with more recent illegal aliens. 4. We assume that all aliens who entered during the same period have the same legal status. In cases where we had assigned illegal status to some aliens and legal status to others who arrived in the same period, we average both their probabilities of being illegal and their random numbers, then assign legal status to the whole entry cohort. (For these calculations, we exclude individuals who are not in the labor force, although they are assigned the same legal status as the rest of their entry cohort.) In cases where, within households, we end up with illegal aliens entering in earlier periods than legal aliens, we assign the same status to all aliens with inconsistent status, using average probabilities of being illegal and average random numbers. 5. On the assumption that legal aliens can usually command substantially higher wages than illegal aliens, in cases where we had assigned a working spouse to be an illegal alien and a nonworking spouse to be a legal alien, we reassign the nonworking spouse to be an illegal alien. 6. We assume that natives, naturalized citizens, and nonimmigrants are able to bring over their spouses through legal means so the alien spouses of these individuals are assigned to be legal aliens. 7. For adult children, we assume that the alien children of natives, naturalized citizens, refugees, and nonimmigrants are legal aliens. 8. For minor children, we assume that the alien children of natives, naturalized citizens, refugees, and nonimmigrants are legal aliens. Of the remaining children, those with at least one illegal alien parent are assigned to be illegal, and all others— that is, those with two legal alien parents or with a single parent who is a legal alien— are assigned to be legal aliens. If an alien minor child is assigned to be a legal alien, we assume that his or her adult siblings who immigrated during the same period are legal as well. For children who are not living with their parents, but are living with other relatives, we derive legal status using the status of the head of their family or subfamily and his or her spouse. For children not living with any relatives, we derive legal status using the status of the head of household and his or her spouse. Estimation procedure: Final legal status adjustments. After applying the consistency checks and making the required adjustments, the number of undocumented aliens in the CPS was about 376,000, roughly 25,000 fewer than the 401,000 we had estimated should be in the March 1995 CPS. This underestimate is not surprising because most of the household and family consistency adjustments resulted in individuals having their status changed from undocumented alien to legal alien. We therefore reran the estimation procedures, but increased the probability that each alien is an illegal alien by 6 percent. With this final adjustment, the number of undocumented aliens in our sample differs by less than 100 from our estimate of the number we expect to find in the CPS. Weighting the March 1995 CPS The March 1995 CPS, a national survey of more than 50,000 households, includes 4,064 households in New York State with 10,532 individuals. The New York sample includes 2,175 foreign-born individuals and 1,141 households headed by immigrants. The sample represents a state population of more than 18.4 million. Thus, each sample case carries a weight (averaging about 1,750) so that sum of the individual weights is equal to the state's population. The assignment of CPS weights is a complex multistage process carried out by the Census Bureau (Waite 1996) to force CPS totals to agree with (a) estimates of the national population by age, sex, and race; (b) estimates of the national population by age, sex, and Hispanic origin; (c) estimates of each state's population aged 16 and over; and (d) for March CPS supplements only, logical requirements that spouses have the same weights. For this project, we modified the official CPS weights in two ways. First, we corrected an error in the CPS weights that led to systematic underestimation of Asians/Pacific Islanders and American Indians/Alaska Natives at the expense of persons of other race; this problem also led to an underestimate of the foreign-born population nationally by more than 800,000 persons (Passel 1996). The second adjustment to the weights was to force the foreign-born population in New York State to agree with the population estimates described in the previous section. Both adjustments are described below. National Weight Adjustment. The March CPSs of 1994 and 1995, as released by the Census Bureau, suffer from a problem with weights, because race was defined inconsistently in the weighting process described in step (a) above. Specifically, in the target population estimates, the population that is not white or black — that is, the "other races" population — is defined as persons who are Asian/Pacific Islander or American Indian/Alaska Native; in the 1994 and 1995 CPS, this other races group also includes individuals who did not specify one of the four races. This inconsistency led to a systematic under-weighting of Asians/Pacific Islanders and American Indians/Alaska Natives by about 20 percent in 1994 and 30 percent in 1995 (Passel 1996). Also, since roughly two-thirds of Asians are foreign-born, the immigrant population is also significantly underestimated by the official CPS data in 1994 and 1995. To correct this inconsistency, we replicated the CPS editing and weighting procedures as closely as possible to develop new CPS weights that more accurately weight immigrants and the rest of the population. The first step in the process involved editing the race item. Persons who did not choose one of the four specific races (that is, chose "other" race) were assigned to one of the four specific races based first on relationships within their households and then, if necessary, by a "hot-deck" process based on the respondent's Hispanic origin. With the race item now consistent with population estimates, the CPS weights were adjusted iteratively to the three sets of population estimates mentioned above. This adjustment process required six iterations. Finally, a process called "spouse equalization" adjusted the weights so that household heads and their spouses or partners had equal weights in each household, while maintaining the age-sex-race/Hispanic origin totals at the national level. (See also Passel and Clark 1997; Passel et al. 1997.) After completing these three steps, we have revised CPS weights that approximate the correct weights for 1994 and 1995.(27) The reweighting has a substantial impact on the CPS estimates of the foreign-born population in both years. For 1994, the foreign-born population of 23,389,000 is 821,000 greater than the Census Bureau's published figure of 22,568,000. For 1995, the difference is even larger— 1,471,000 more than the published figure of 22,960,000. The revised weights are used throughout this analysis. Foreign-Born Weighting. After adjusting the CPS weights and assigning legal status to the individual CPS cases, we found that the foreign-born population of New York was 3.1 million. For some components of the foreign-born population, our estimates differ from the CPS figures. The revised CPS shows 246,000 refugees while we estimate there are 198,000. For naturalized citizens, the CPS has 1,072,000 while we estimate there are 1,187,000.(28) For the remaining group of legal permanent resident aliens plus undocumented aliens, the CPS has 1,820,000 while we estimate there are 1,937,000. Although these differences are actually not particularly large in percentage terms, when the estimates are disaggregated by region of origin and period of entry, some discrepancies are fairly large. We introduce a final weighting adjustment to bring the CPS figures for the foreign-born population into closer agreement with our population estimates. The adjustments consist of proportionally adjusting the CPS weights within period of entry, country of birth, and immigrant status groups so that the sum of the adjusted weights equals the population totals. In order to avoid giving undue weight to a small number of cases, we require that each country grouping used for weighting have at least 10 CPS cases for refugees and naturalized citizens and at least 25 for aliens. For refugees, we use two periods of entry— 19901995 and 19801989; for aliens and naturalized citizens, we use the same two periods plus the earlier, pre-1980 arrivals. The country groups used for this adjustment are shown in Table N.
The final weighting adjustment is for the remaining group of aliens, that is, those who are not naturalized citizens, refugees, or nonimmigrants. First, we calculate the CPS weighted totals for all aliens who are not nonimmigrants or refugees to provide the denominators of the adjustment factors. The numerators are the estimated number of nonrefugee, legal permanent residents plus the estimated number of undocumented aliens minus the number of nonrefugee naturalized citizens based on the final weights from the previous step. These ratios times the CPS weights for individual aliens give the final weights for aliens. For nonimmigrants, we use the CPS (as adjusted in the previous national procedure) as the final weights. Table 2 shows the foreign-born population by country of birth and citizenship that results from the application of the classification and weighting procedures. For U.S. natives, the CPS weights used are those from the national reweighting procedure. The second generation is defined as U.S. natives with one or two foreign-born parents. Third-and-higher-generation U.S. natives are those who have two U.S.-native parents. It should be noted that U.S. natives include persons born in Puerto Rico, the U.S. Virgin Islands, and other U.S. territories, and persons born abroad to U.S. parents. The populations of New York City and the rest of the state are not controlled in the weighting process; rather the populations of the two parts of the state arise as the sum of weights of CPS respondents. V. Tax EstimatesOverview of Estimation Strategy Our initial estimates of income and taxes paid by each household are made by using the Urban Institute's TRIM2 (the TRansfer Income Model, Version 2), a microsimulation program that estimates tax liabilities by applying extremely detailed tax rules to individuals and families in the March supplement to the CPS. For example, to estimate federal income tax, TRIM2 fills out a tax form for each tax filing unit in the CPS sample (Giannarelli 1992).(29) TRIM2 is used to estimate federal and state income taxes, FICA payments, and unemployment insurance payments. Although TRIM2 imputes estimates of capital gains, we found a significant understatement of income and income taxes at the top of the income distribution, a well-known property of the CPS and TRIM2. To compensate, we devised an imputation procedure to add additional income to high income households. We used these adjusted incomes in our other tax estimates, which then generally needed only small further adjustments, suggesting that the adjustments for federal income tax made the income data more accurate. We also estimate three other taxes— New York City income tax, state and local general sales tax, and residential property tax. New York City income tax is estimated as a percentage of New York state income tax. Thus, we derive estimates of the city income tax from the TRIM2 estimates of State income tax. The estimates of residential property tax are based on the 1996 New York City Housing and Vacancy Survey, the March 1995 CPS, and administrative data. To estimate sales tax, we use a very simple household consumption model and administrative data. In this section, we describe the tax estimation methodology in more detail, with particular attention to adjustments to the TRIM2 output. Where possible, we compare our estimates with administrative data on taxes collected. Often the categories available for comparison do not correspond closely to the data we have available. However, to the extent possible, we use the administrative data to assess the quality of our estimates and to make further adjustments to the estimates where necessary. We also discuss the resulting estimates of income and tax payments. Federal Income Tax TRIM2 estimates federal income tax liability for tax filing units. The program divides households into tax units, each consisting of an individual or married couple, along with their dependents, and determines who is required to file a tax return. (In most cases, the tax unit corresponds with the household.) Tax liability estimates are based on wage and salary income, self-employment income, farm income, interest, rents, government and private pensions, dividends, capital gains, Social Security income, and unemployment insurance. Information on capital gains and most itemized deductions is not collected in the CPS; these items are imputed using a statistical matching procedure based on federal income tax returns. A detailed description of TRIM2's estimation methods can be found in Giannarelli (1992:153171). The TRIM2 simulation programs apply detailed federal tax rules to each family in the CPS microdata file. In effect, the computer program fills out the tax forms for each family. TRIM2 counts dependents, adds up income, subtracts adjustments to income, subtracts the larger of itemized or standardized deductions, subtracts personal exemptions, computes taxes on taxable income, and computes and subtracts tax credits to arrive at final tax liability (Giannarelli 1992:4). It should be noted that TRIM2 calculates tax liability, not taxes paid. The difference may, in part, represent the degree of compliance with the tax code; as we describe later, we make an explicit allowance for noncompliance by undocumented aliens. Initial Federal Tax Estimates. The initial TRIM2 federal tax estimates are shown in Table 5, together with information from the IRS' Statistics of Income (SOI) for tax year 1994, that is, tax returns filed in 1995 for income earned in 1994 (Internal Revenue Service 1997). To make the best comparison of the two data sources, we use data from the IRS on tax returns filed that have some tax liability; from TRIM2, we display data for tax units that have income other than welfare and "other" income and that are estimated to have some tax liability. In terms of the number of returns, the simulated data are close to the SOI administrative data: 6,799,000 tax units versus 6,545,000 returns— a difference of 254,000 or 3.9 percent. For Adjusted Gross Income (AGI) and tax liability, the simulated results show severe departures from the IRS data. AGI is understated by $26.5 billion or 9 percent of the $300 billion total. Tax liability misses the target by even more, falling short by $9.7 billion or 21 percent of the $45 billion collected from New York State. The differences in the two sets of estimates point to the upper and lower ends of the income distribution as the source of the overall discrepancy. For the lowest category of returns (AGI less than $15,000), the TRIM2 simulation generates the same number of returns, but understates aggregate AGI significantly— by $2.2 billion or 15 percent. Further, the TRIM2 simulation shows a negative tax liability (from the Earned Income Tax Credit or EITC) for this category as a whole. This pattern suggests that the TRIM2 simulation generates too many EITC returns, perhaps because not all persons who are eligible for the EITC actually file returns or because households understate their incomes in the CPS, and therefore are erroneously classified in the lowest income category. Regardless of the explanation, the understatement of income and tax payments from this low income group has little impact on the overall results of our analysis. The $2.2 billion shortfall in AGI, while clearly a large sum of money, represents only about 0.7 percent of the $300 billion in AGI reported on federal tax returns filed from New York. The shortfall in estimated taxes, $235 million, is an even smaller percentage— 0.5 percent— of the more than $45 billion collected from the state. The difficulties faced by the TRIM2 simulation at the upper end of the income distribution are of much greater significance for the overall accuracy of the tax and income estimates. The highest income category, AGI of $200,000 or more, represents only 108,000 tax returns according to the IRS, or 1.6 percent of the returns with any tax liability, but these filers report an astounding $59 billion in AGI, or 19.7 percent of the AGI. They pay an even larger percentage of taxes, 37.0 percent or $16.8 billion. TRIM2 clearly has a problem with this category, It simulates only 39,000 returns with AGI of $9.9 billion and a tax liability of $2.6 billion. Thus, TRIM2 only accounts for 36 percent of the highest income returns, only 17 percent of their AGI, and 15 percent of their tax liability. Three factors account for the understatement of this very high income group in the CPS and TRIM2. First, their relative rarity makes them unlikely to fall into the CPS sample. Second, very high income households may be extremely difficult to interview, both in terms of access to secure dwellings and refusal to be interviewed. After all, how likely is it that Bill Gates, Ted Turner, or Michael Jordan would consent to a CPS interview even if they did fall into the sample? The third factor is underreporting of income combined with top-coding in the CPS data file. This third factor is readily apparent in the mean AGI information. The mean AGI in the top group of IRS returns is $548,000, whereas the TRIM2 simulated group has a mean AGI of less than half this value, $257,000. The two income categories just below the highest group, $75,00099,999 and $100,000199,999, are affected by the same factors. At first glance, the number of tax units in these groups appear to be overreported, not underreported. Each group has more returns in the simulation than in the actual IRS data— 114,000 or 31 percent more for the $75,00099,999 returns and 126,000 or 43 percent more for the $100,000199,999 returns. Since the simulated mean AGIs and tax liabilities differ little from the reported data — no more than 3 percent for either group — the simulated aggregate AGIs and tax liabilities for these groups overestimate the IRS reported totals by percentages similar to the overstatement of returns. Thus, it would appear simply that TRIM2 and the CPS "overcover" these income categories. However, the properties of the income distribution and the fact that each of these groups has an upper income limit argue strongly for underreporting of income. If income is systematically underreported in the CPS by households with simulated AGIs in the $75,00099,999 range, then the true AGIs of the households would place many of them in higher income categories. Thus, correcting the income underreporting in these higher incomes would bring the simulated returns for the $75,00099,999 category into closer agreement with the IRS data. Further, the aggregate AGI and tax liability for the category would shrink as households moved into higher categories; the means, however, could well remain virtually unchanged if the incomes of those remaining below the upper limit of the category also saw their incomes rise. With a correction for underreporting of income in the $75,00099,999 category, many of these households would shift into the $100,000199,999 category, thus exacerbating the overreporting of the number of tax units in this group. However, some of the households in this category should probably fall into the top AGI category, $200,000+. To adequately measure taxes paid by different segments of the population, it is necessary to correct the TRIM2 simulation for the systematic understatement of income and taxes in the highest income households. We report in the next section on the implementation of methods for adjusting the CPS and TRIM2 data. One unsolved issue is that TRIM reports more returns for tax units with AGI of at least $75,000 than the IRS does. For all returns with AGI in excess of $75,000, TRIM2 shows 943,000 returns or 22 percent more than the IRS report of 772,000 returns. We have not investigated this phenomenon in detail, but it may be possible that the simulation fails to disaggregate some returns from two-or-more-earner households into multiple returns that would fall below the highest income categories. Our adjustments, reported below, focus on accurately simulating income and taxes, not returns and exemptions.(30) Correction for Underreporting of Income and Taxes. From the foregoing discussion and the initial tax estimates, it is clear that AGI is underreported for upper income tax filers. We adjust the data by adding income to higher income returns. However, several different methods are available for doing so. One alternative is simply to multiply all higher incomes by some adjustment factor or add a constant dollar amount to each return. However, corrections of this nature leave "holes" in the income distribution and fail to extend the high end of the income distribution enough. For example, if we added $5,000 AGI to every return with incomes over $75,000, then there would be no returns with AGI between $75,000 and $80,000. The same phenomenon occurs if we multiply all incomes by a common factor. Our initial income adjustment strategy circumvents these problems by only increasing that portion of AGI above a certain threshold. For returns with AGIs in the $75,00099,999 range, we add additional AGI to each return equal to 25 percent of its AGI above $75,000. For those with $100,000199,999 AGI, the additional AGI is equal to 25 percent of that in excess of $100,000; and for $200,000+, we add 25 percent of the AGI above $200,000. Additional tax for each household is calculated by applying the appropriate marginal tax rates to the additional, adjusted income. This first adjustment, shown in Table 5 as "Increase over $75,000," is designed to focus on those returns at the lower end of the upper range, while preserving the overall shape of the income distribution by not leaving "holes."(31) With this adjustment, the total AGI is increased by almost $4.9 billion to $278.3 billion; this new figure still falls $21.6 billion or 7.2 percent short of the official total AGI on tax returns of $299.9 billion. However, the adjustment brings the upper end of the income distribution somewhat closer to the SOI data. At $75,00099,999, the estimated number of returns is reduced by 83,000 and is now only 31,000 or 8.3 percent, too high. For this category, the average AGI and average tax liability differ very little from the official SOI figures. In the top income category ($200,000 AGI and over), the first adjustment adds 36,000 returns and cuts the shortfall of returns in half. For the next AGI category, $100,000199,999, the average AGI and tax now differ by less than 5 percent from the official figures. With this adjustment, virtually the entire shortfall in AGI and tax liability is accounted for by these top two income categories. These two miss the official AGI estimate by $21.7 billion while the aggregate shortfall is $21.6 billion; for taxes, the two categories fall short by $8.3 billion while the total misses by $8.1 billion. From these new adjusted figures, it is clear that the top AGI category is still missing considerable income (and tax liability). Even with the adjustment, the average AGI for those returns over $200,000 is only $240,000, whereas the returns filed with the IRS in this category have an average AGI more than twice as large, $547,000. At this point, we decided to adjust only the top income category to make up for the missing income and tax liability. The procedure used is similar to the first adjustment. For this final adjustment, only AGI in excess of $200,000 per return is increased; the factor used to bring the adjusted distribution into agreement with the IRS figures is 6.82. Again, the tax liability on the additional adjusted income is determined by applying the appropriate marginal tax rates. With this final adjustment, the total adjusted gross income on simulated federal income tax returns for New York State is $300.1 billion, a difference of only $295 million, or 0.1 percent, from the $299.9 billion shown in official IRS data. The estimated tax liability is $45.9 billion or 1.3 percent different from the official figures. The top two income categories have offsetting estimates in that our data overestimate AGI for the $100,000199,999 category by $19.4 billion, but underestimate the liability in the $200,000-and-over category by $19.1 billion. Given the limitations of CPS data for very high income households, we have opted to keep these estimates in this form and not try to fix all of the departures from the official data. At this point, the average AGI and tax liabilities give a reasonable approximation to the SOI data; for example, the average AGI for returns with incomes in excess of $200,000 is now $534,000, or less than 3 percent below the official data. Allocation of Income and Tax Estimates to Household Members. The income and tax model just described estimates the AGI and federal income tax liability for tax units (which, in most cases, correspond to households). However, our goals are to estimate total income and to allocate all income and taxes to individuals, not households. Towards this end, we first assume that the additional AGI added to higher income households also represents an addition to total income.(32) Then, we apportion the additional income among household members in proportion to each person's (nonwelfare) income. Next, for all households, we apportion the federal income tax in proportion to the total adjusted income from all sources except welfare. We use the same assumption to apportion state income tax and New York City income tax. Since the sales tax model is based on consumption, for general sales tax, we use all sources of income, including welfare. FICA and unemployment insurance are estimated directly for each person by TRIM2, so no allocation procedure is required. For property tax, a somewhat different method, described in the property tax section, is used. TRIM2 estimates tax liability, not tax payments; the mediating factor between the two is taxpayer compliance. In general, compliance levels are high in the United States and, in any case, the effective compliance rate implied by the TRIM2 estimates is 100 percent because the TRIM2 estimates, in general, are calibrated to actual tax collections. The one main group for whom this assumption may be off the mark is undocumented aliens. Given that many undocumented aliens do not have papers, it is undoubtedly the case that many work "off the books" and so do not pay income taxes on their earnings. On the other hand, many are working with fraudulent documents or "borrowed" documents. In such cases, it may not be possible for the undocumented workers to recover taxes withheld and they may overpay income taxes. There is very little information on tax payments by undocumented aliens. In a 20-year-old study, North and Houstoun (1976) found that almost half of undocumented workers were having taxes withheld. More recently, a small survey in San Diego found that about half of undocumented workers had taxes withheld from their pay (Rea and Parker 1992). Based on this information, we assumed a compliance level of 60 percent for undocumented workers in the CPS. To implement this assumption, we simply multiplied tax liability by 0.60 for every undocumented alien in the CPS, rather than assigning either full compliance or zero compliance randomly. This assumption applied to the three income taxes, FICA, and unemployment insurance. With this implementation, the effective compliance rate is actually lower, however, because not all undocumented workers are represented in our CPS data set. For sales and residential property tax, we assumed full compliance. Adjusted Income Estimates. Total aggregate personal income in New York State is $330.2 billion. Immigrants account for $57.5 billion or 17.4 percent of the total and slightly less than their share of the population, 18.2 percent (Table C, page 8). Households with incomes of $200,000 or more account for a very large share of total aggregate income in the state. These households constitute 1.3 percent of the households in the state, but are responsible for 13.1 percent of the total aggregate income (Table S, page 59).
Two factors explain the differences between the per capita and the household-level results. First, a higher percentage of natives than legal foreign-born are children who have low incomes and tax contributions. Among natives, 29 percent are under 18 years old, while only 10 percent of the legal foreign-born fall in this age group. In the individual level estimates, the relatively high proportion of children lowers the estimates for natives relative to the legal foreign-born. In the household-based estimates, the low income and tax contributions of U.S.-born children of immigrants are attributed to the legal foreign-born— their parents— rather than to natives. All together, 34 percent of households headed by a legally present immigrant contain a native child. Second, although on average households headed by the legal foreign-born contain slightly more adults (aged 18 and older) than households headed by natives, 2.0 versus 1.8, the average incomes of adults in households headed by the legal foreign-born are substantially lower than the incomes of adults in households headed by natives, $18,400 versus $25,500. Natives and the legal foreign-born are about equally likely to head extremely wealthy households, those with total income of $200,000, 1.3 percent versus 1.2 percent, respectively, but among wealthy households the native-headed households have substantially higher average incomes. Whether measured at the individual or household level, the incomes of undocumented aliens are substantially lower than those of both natives and the legal foreign-born. Per capita income for the legal foreign-born is $18,000, compared with $12,100 for undocumented aliens. Average total household income for households headed by the legal foreign-born is $38,700, compared with $32,400 for households headed by undocumented aliens. Although households headed by undocumented aliens contain slightly more adults, on average, than those headed by the legal foreign-born, 2.1 versus 2.0, the average incomes of adults in households headed by the undocumented aliens are substantially less than those of adults in households headed by the legal foreign-born, $13,700 versus $18,400.
Besides having substantially lower incomes than other legally present immigrant groups, refugees also are substantially more dependent on welfare (Supplemental Security Income, Aid to Families with Dependent Children, and other public assistance). Forty-one percent of households headed by refugees receive welfare compared with 8 percent for naturalized citizens, 24 percent for LPR aliens, and none for nonimmigrants. Twenty percent of income received by households headed by refugees comes from welfare, compared with 1 percent for naturalized citizens and 4 percent for LPR aliens.
In New York State, 74 percent of the legal foreign-born live in New York City, so the characteristics of the immigrants in New York City dominate statewide averages. In fact, outside New York City, the average incomes of the legal foreign-born equal or surpass those of natives, no matter how they are measured. (See Table F, page 23.) Outside New York City, at the individual level, the average income of the legal foreign-born, $23,900, surpasses that of natives, $19,100. Per capita incomes for naturalized citizens and LPR aliens are particularly high, $30,000 and $20,400, respectively. Unlike the state overall, outside New York City, the average incomes of households headed by the legal foreign-born surpass those of natives. For the legal foreign-born, average household income is $54,300, compared with $52,700 for natives. Average incomes for households headed by naturalized citizens, $54,900, and LPR aliens, $54,900, are both greater than that of natives.
Federal Income Tax Estimates. Natives in New York pay $39.1 billion in federal income tax or 85.5 percent of the total paid by New York residents. Since natives represent 81.8 percent of the state's population, natives pay a higher share of taxes than they represent in the population (Table R). Legally present immigrants, representing 15.7 percent of the state's population, pay $6.4 billion in federal income taxes, or 13.9 percent of the total paid from New York. However, naturalized citizens pay a greater share of the tax, 7.8 percent, than they represent in the population, 6.3 percent. Refugees and undocumented aliens pay significantly smaller shares of income tax. For refugees, who are 1.1 percent of the population, the 0.2 percent they pay in taxes is so small because of their low incomes and the large amount of nontaxable welfare income they receive. Undocumented immigrants pay only a small share of income tax because of low incomes and low levels of compliance. The upper end of the income distribution accounts for a disproportionate share of income and taxes. Households with income of $200,000 or more represent only 1.2 percent of all households in New York, but have 12.4 percent of all income (Table S). These households pay an even higher percentage of the federal income tax— 30.4 percent or $13.9 billion. There are 87,000 households in this high income category, with a total of $41 billion in personal income. The average household has an income of $470,000 and pays a federal income tax of $160,000.
This progressivity affects the balance of federal tax payments between natives and immigrants through this highest income group. While natives and legally present immigrants have approximately the same proportion of households with incomes of $200,000 or more— 1.3 percent for natives versus 1.2 percent for legally present immigrants — the native households have much higher incomes on average ($512,000) than the legally present immigrant households ($280,000). As a result, the percentage of federal income tax paid by natives is higher relative to their representation in the population than for legally present immigrants (Table R, page 58). The difference is not great, however, from the perspective of the average household. On average, natives pay 14.4 percent of their income in federal income tax while legally present immigrants pay 12.3 percent. The federal income tax is more progressive than any of the other taxes we analyzed, as indicated by the percentage of tax paid by high income households in Table R. The New York State income tax and New York City income tax are also more progressive than the other taxes, but less so than the federal income tax. The progressivity can be gauged by the tax rate on high income and by the income at which the highest rate begins. For husband-wife households, the highest tax rate for New York State is 7.875 percent for income exceeding $26,000. For New York City the highest rate, 4.46 percent, begins at an adjusted gross income of $108,000. (New York City, however, has a smaller percentage of very high income households. See below.) New York State Income Tax The method for estimating New York State income tax is very similar to the one used to estimate federal income tax. TRIM2 first applies New York State tax rules to the CPS households and estimates tax liability for each household. The New York State Department of Taxation and Finance (1997a) publishes aggregate information on state income tax returns. They define three populations: full-year residents, nonresidents, and part-year residents (divided into in-movers and out-movers). The CPS provides information on current residents, so it is necessary to choose a reference population in the administrative data to compare with our CPS estimates. We assume that all CPS households are full-year residents. By making this assumption, we erroneously impute income earned out-of-state by in-movers as New York State income. In the same manner, we miss income earned in New York State by out-movers. Overall, both of these figures are small and offset one another to a large degree. For 1993, the potential error introduced by this assumption amounted to $60 million in tax liability, or less than 0.5 percent of tax collections. Table 6, which compares the TRIM2 estimates to the New York State administrative data, shows patterns similar to those observed for federal income tax in the unadjusted estimates. First, TRIM2 underestimates the number of returns in the lowest AGI bracket, under $10,000. However, this is principally a definitional issue since the state data include returns with no tax liability while the TRIM2 data do not. In fact, the total tax liability for this income bracket is misestimated by only $14 million, or only 0.1 percent of the total liability of more than $13 billion. For tax units in the highest bracket, TRIM2 seriously understates the amount of income and taxes, so we added the additional AGI from the federal tax model and estimated additional New York State income tax liability on the basis of the marginal tax rates. Overall, the AGI adjustments made in the federal tax estimation lead to consistent, but small, overestimation of New York State AGI for the TRIM2 model. The general pattern suggests that TRIM2 overestimates AGI in all brackets and thus consistently overestimates income tax liability. After both AGI adjustments (the first for all households with AGI over $75,000 and the second for those over $200,000), the aggregate AGI exceeds the published figure by 9.8 percent, with excesses at all income ranges (not shown in Table 5). Differences in how New York and the federal government define AGI undoubtedly account for a large share of the discrepancy. To bring the two sets of tax estimates into line, we multiply tax liability for all households with AGI in excess of $10,000 by 0.91. With this final adjustment, the aggregate New York State income tax liability is estimated at $13.5 billion, only $170 million or 1.3 percent higher than the state's administrative total. The mean tax liabilities and mean AGIs for income categories are all very close to the administrative totals, with the exception of the highest category. For returns with AGI in excess of $100,000, the TRIM2 simulation estimates too many returns (190,000 more than the 352,000 actually filed), but understates the average income and tax liability. The result is that the taxes paid are estimated within 0.8 percent or $48 million out of $5.9 billion.
The state income tax is somewhat progressive. Households with incomes of $200,000 or more pay 21.1 percent of the state income tax, even though they represent only 1.2 percent of the state's population (Table S, page 59). This progressivity and the underrepresentation of immigrants in the very high income households account for a large part of the somewhat low payments by immigrants. However, the proportion of household income paid in state income tax is not very different for native-headed households (4.2 percent) and households headed by legally present immigrants (3.8 percent). New York City Income Tax The method for estimating New York City income tax is a straightforward extension of the state and federal income taxes. New York City income tax is a direct function of the state income tax for income earned in New York City. Individuals are liable for New York City income tax if they work in the city, not just if they live in the city. Unfortunately, the CPS provides information on residence and income, but not on place of work. Thus, we are not able to identify the population earning income in New York City, but only the population living there. Accordingly, we estimate New York City income tax for persons living in New York City.(33) Some of the tax due from nonresidents will be offset by the imputed tax on income of city residents earned outside the city. We develop a set of factors to derive New York City income tax from New York State income tax. To do so, we first compute the tax liability for selected income values and then determine the average factors for different brackets. A simple model with two brackets fits the data quite well. For AGIs in the upper bracket, estimated New York City tax is 0.565 times the unit's state tax; in the lower bracket, the multiplier is 0.60. The AGI cutoff varies by filing status: for single filers, $12,500; for joint filers, $30,000; and for heads of households, $25,000. Table 7 compares our TRIM2 simulated New York City tax estimate with data from the City's Department of Finance (1997a) for returns filed for tax year 1994. The total number of returns is similar for the two estimates, differing by 31,000 or 1.4 percent. However, the simulation seriously understates the number of low income (AGI less than $15,000) returns and overstates the number of high income (AGI greater than $100,000) returns. The shortfall in low income returns derives from the similar shortfall in state tax returns (Table 6), but again has little impact on the overall estimates of taxes collected. The difference in tax estimates for low income returns amounts to $9 million or about 0.3 percent of the $2.9 billion in taxes collected. For returns with AGI greater than $100,000, the TRIM2 estimates seriously understate the average AGI ($162,000 versus $292,000) and the mean tax liability ($5,400 versus $11,500). Even with the compensation of a higher number of such returns, the simulated estimates show that taxes owed by this high income group ($1.037 billion) fall $274 million, or 21 percent, short of the official figure of $1.312 billion. This shortfall accounts for about 75 percent of the overall underestimate. There seem to be two possible sources for this underestimate. First, on high income returns, there are differences in how the city, state, and federal tax systems define AGI that are not fully captured by the simulation/estimation process. Second, our simulation will miss income and taxes attributable to nonresidents of the city. Some of this "missing" tax is attributable to nonresidents of New York State and some to New York residents living outside the city and we suspect that high income households are especially likely to be missed. Accordingly, we make no further adjustments to the city tax estimates.
New York City Income Tax Estimates. Although the New York City tax rate schedule is more progressive than the state's, the city income tax is effectively less progressive than either the state or federal income tax. High income households, with incomes exceeding $200,000, pay 17.4 percent of the tax (Table S, page 59) while accounting for 9.8 percent of the personal income of New York City residents. The lower effective degree of progressivity stems principally from the fact that there are proportionately fewer of the very high income households in New York City than in the rest of the state. As a result, natives pay 2.3 percent of their income in New York City income tax while LPR aliens and naturalized citizens pay 2.1 percent. The other immigrant groups all pay a lower percentage (Table U). Residential Property Tax We estimate residential property tax differently for New York City than for the rest of the state because there are far more data available on residential property tax for New York City than for the rest of the state. The data available for New York City and the rest of the state vary in three significant ways. First, for most of the state, data are not available for residences in commercial buildings— in other words, most apartments; tax assessments for these units cannot be readily distinguished from tax assessments for other commercial real estate. An exception is condominiums, which, regardless of their location, are included in the "residential" assessments. In New York City, on the other hand, assessments for all residential real estate, including commercial apartment buildings, are available. Second, the 1996 New York City Housing and Vacancy Survey (NYCHVS) provides detailed housing characteristics of city residents, including the amount they report paying in property tax, as well as information about whether the head of the housing unit was born in the United States. As described below, these data allowed us to directly model the property taxes paid by foreign-born and native household heads. No comparable data are available for the rest of the state, so we had to rely on stronger assumptions to estimate property taxes paid by immigrant groups and natives. Third, the years for which property tax assessments are available differed for New York City and the rest of the state. For the state as a whole, the latest assessments we could obtain are for 1994 and are based on a 1993 market value standard (Office of Real Property Services 1997). For New York City, the residential property tax assessments refer to fiscal year 1997 (Department of Finance 1997b). Since, in both cases, we had only a single year of data, we did not attempt to make adjustments. Because of the differences in data available, we develop somewhat different methodologies for calculating residential property taxes. However, for both sets of estimates, we make some of the same basic assumptions:
In the descriptions below, we explain how we calculate credit for property taxes attributed to individuals (that is, the head, his or her spouse if there is one, and any household member who is a landlord, regardless of whether he or she is the head or spouse). We also produced a second set of estimates in which credit for property taxes is attributed to the head of household. Household estimates are calculated as the sum of the individual estimates. Residential Property Tax for New York City: Households headed by natives and the foreign-born from the NYCHVS. We use data from the NYCHVS— a survey of 15,752 households conducted by the Bureau of the Census— to make baseline estimates of residential property taxes paid by New York City inhabitants. The estimation procedure has three steps: (1) imputing residential property tax for homeowners for whom residential property tax is missing; (2) imputing residential property tax for renters; and (3) adjusting property taxes to conform with administrative data on tax assessments. (1) Imputing property taxes for homeowners with missing information. About two-thirds of all owned housing units are missing information on residential property tax. Using the households for which property tax is reported, we model property tax using the following variables: reported value of housing unit, total household income, year built, number of units in the building, presence of mice, whether buildings in the neighborhood have broken windows, whether the building is boarded up, whether the unit is a condo, borough, whether the sub-borough is "high rent" (that is, with mean rent higher than at least 80 percent of the sub-boroughs in the city), and whether the household head is black. The most important predictor of property tax is the reported value of the housing unit. The reported value for most coops appeared to be incorrect— 60 percent had a reported value below $100,000 and 20 percent had a value below $20,000. We excluded all coops from our models of residential property tax. The values of all coops were imputed, using the model described in the next section. (2) Imputing property tax for rental units. The first step in imputing property tax for rental units is to impute the value of the dwelling, which is not reported. The model estimating the unit's value uses information on owned units and contains the following variables: total household income, year built, number of units in the building, number of rooms, number of bedrooms, whether the unit is a condo, whether buildings in the neighborhood have broken windows, an index of neighborhood quality, borough, whether the sub-borough is high rent, and whether the household head is black. We estimate the property tax paid on the rented dwelling with the model used to impute property taxes for households in owned units for which property tax is missing, described in the previous section. The imputed value of the dwelling unit is used in the property tax imputations. Because $100199 is the lowest value of real estate taxes reported in the survey, we set a floor for imputed property tax of $150; imputed values below $150 are recoded as $150. (3) Adjusting property taxes to conform with administrative data on tax assessments. The aggregate estimated residential property tax contributions exceed official estimates by about a third. In order to make our estimates conform with administrative estimates, we reduce our estimates of property taxes paid on rental units (both the renters' and landlords' components) by approximately 38 percent. Individual-level estimates from the CPS. The NYCHVS analysis produces estimates of the aggregate property tax for five groups:
To apportion the estimated property tax to individuals within each group of owners and renters,
we assume that residential property tax contributions are proportional to household income.
Therefore, property taxes contributed by individual natives in owned homes and their spouses
(
If the head has no spouse, estimated taxes contributed by that head are
To estimate residential property tax attributable to individual landlords,
Estimating Residential Property Tax Outside New York City. In estimating
residential property tax for the rest of the state, we had less information than we had for New
York City. All we had was total residential property tax and, as mentioned, even this figure
excludes contributions made for units in commercial apartment buildings. Our underlying
assumption is that property tax contributions for a dwelling are proportionate to the household
incomes of the dwelling's inhabitants. Estimated total tax contributions outside New York City(36)
for owners (
The estimated residential property tax of an individual owner with a spouse, or the spouse, is:
 .
For a head with no spouse, the estimated contribution is
To estimate residential property tax attributable to individual landlords,
Residential Property Tax Estimates. Table 8 shows the results from the property tax estimation model. Overall in New York State, the shares of residential property tax that natives and the legal foreign-born pay are proportionate to their shares of the state population: natives constitute 82 percent of the population and pay 83 percent of the residential property tax, and the legal foreign-born constitute 16 percent of the population and pay 17 percent of the property tax (Table V). Undocumented aliens, who constitute 3 percent of the state population, pay 1 percent of the residential property tax.
The legal foreign-born, constituting 16 percent of New York State's population, account for 15 percent of the residential property tax paid by owners and renters, while natives, constituting 83 percent of the state's population, account for 85 percent (Table 8). There are two major reasons the residential property tax contributions as owners and renters of the legal foreign-born are somewhat lower than those of natives, which also explain the low property tax contributions of undocumented aliens. First, residential property tax contributions of a large share of the foreign-born outside New York City are not included in our estimates. As mentioned, this is because of lack of data on property taxes paid on the commercial apartment buildings outside New York City. This omission disproportionately affects the foreign-born, particularly undocumented aliens. According to our estimates, outside New York City, only 5 percent of natives live in commercial apartment buildings, compared with 15 percent of the legal foreign-born and 30 percent of undocumented aliens. Thus, the relative tax contributions of the foreign-born, especially undocumented aliens, are underestimated. Second, foreign-born individuals, particularly undocumented aliens, are substantially less likely to own their own homes, and more likely to rent their dwellings, than natives. In our residential property tax estimates, we credit home owners with all property tax paid, but renters with only half of the property tax paid on the unit. Statewide, 62 percent of natives live in owned units and 29 percent in rental units for which residential property tax could be calculated.(37) For the legal foreign-born, 33 percent are in owned units and 59 percent in rental units for which residential property tax could be calculated. For undocumented aliens, 11 percent are in owned units and 81 percent in rental units for which residential property tax could be calculated. The legal foreign-born account for 29 percent of the state's residential property taxes paid by landlords, although they only constitute 16 percent of the state's population (Table 10). In contrast, natives, constituting 82 percent of the population, pay only 70 percent of the residential property taxes paid by landlords. The relatively high residential property tax contributions of landlords among the legal foreign-born is due to naturalized citizens who entered the United States before 1980. Although this group constitutes only 5 percent of the state's population, they are responsible for 24 percent of the residential property taxes paid by landlords.
General Sales Tax — State and Local General sales tax is estimated as the sales tax rate times the estimated amount of money a household spends on taxable items. Spendable income is estimated as the household's after tax income less housing and food costs. In addition, we make an allowance for immigrants to remit money to their home countries for relatives there. The model provides a first approximation for each household's sales tax expenditures. The total amount for the state estimated from the model is scaled to agree with the amount actually collected according to the state's administrative figures with a factor that allows for household spending on nontaxable items. Finally, the estimate of sales tax from a household is allocated to the individual household members in proportion to their incomes. Housing Expenditures. Money spent on housing is not subject to New York State or local general sales tax, nor is it available to be spent on taxable items. Housing expenditures are a function of a household's income, the type of dwelling, and possibly other characteristics of the household. Unfortunately for our purposes, the CPS provides almost no information on a household's spending on housing. However, CPS collected some data on the characteristics of each household's housing and a great deal of data on the household's socioeconomic characteristics. We investigated the determinants of housing expenditures using data from the 1996 New York City Housing and Vacancy Survey (NYCHVS). Our strategy is to model housing costs as a function of the household's characteristics. Because we ultimately must estimate housing costs for CPS households, the choice of variables to use in the model is limited to those variables available in the CPS: tenure (owner/renter), units in structure, and household and personal socioeconomic characteristics. We investigated several functional forms— modeling the percentage of income spent on housing, directly or in logit form; modeling the amount spent on housing— before settling on a model of percentages in logit form. Tenure was determined to be a significant factor in the percentage of income spent on housing. Because tenure affected the nature of the relationship between housing costs and the determinants, we opted for separate estimating equations rather than a single equation including a dummy variable for tenure. Nativity of the household head was not a significant factor in determining housing costs. Housing costs in the NYCHVS are the sum of mortgage payments, rent, utility costs, property tax, insurance, and condominium fees. The regression equations for the percentage of a household's income spent on housing costs are estimated separately for owners and renters. The equation for owners is
 = percent of total income spent on housing;
The regression equation for renters is
 = 1, if a single-family dwelling; = 0, otherwise;
the other variables are defined as in the previous equation. The appropriate equation is calculated for each household in the March 1995 CPS, and the resulting value of the percent of income spent on housing is used in the sales tax estimating equation (below). Household Food Cost. In New York State, groceries are not taxed, so it is necessary to take into account food costs in determining how much of a household's income is spent on taxable items. We used the 199495 Consumer Expenditure Survey (CES) to measure food costs. This survey collects detailed data on household spending patterns. The data cover essentially all of an average household's expenses, including food at home, food away from home, alcoholic beverages, shelter costs, utilities, housekeeping supplies, household operations, furnishings, apparel and services, transportation, entertainment, pensions and insurance, personal taxes, health care, education, donations, gifts, and various other expenses. We used tabular data from the CES posted on the Bureau of Labor Statistics (1997) website in our estimation process. Tabulations are available on household expenditures for income groups by age of householder, by region, and by household size; for single persons by sex, tabulations are available by sex and by income. For each category, mean expenditures and income are provided as well as information on selected other household characteristics. The dependent variable for our analysis is the amount spent on food at home per person in the household. For each group (that is, income-age, income-region, income-size, sex-age, sex-income), we computed this value as the average amount spent on food at home divided by the average household size. We then fit the following model to these group means:
 = food costs per person; and
the other variables are defined as in the previous equations. Food cost per person is then
calculated from this equation for each household in the March 1995 CPS. The total amount
spent by the household for food at home,
Taxable Spending and Sales Tax Estimates. With the estimation of housing and food costs, we have made allowance for the two main expenditures that do not generate sales tax attributable to the household. There are two other factors to be taken into account in our household expenditure model. Immigrants often send remittances back to their home countries to provide support and investment income for relatives who remain behind (Massey et al. 1994). In our model, we allow for each adult immigrant with earnings of $5,000 or more to remit $1,000. The remittance money does not generate any sales tax revenue for the state because it is not spent in the United States. Finally, we allow for the fact that not all spendable income is spent on goods and services that are taxable; examples would include savings, gambling, some entertainment, and certain services. We model this as a multiplicative factor that reduces taxable spending. This factor permits us to scale the aggregate sales tax revenue collected to agree with administrative totals. Thus, the model for taxable spending is
 = spending on items subject to sales tax (that is, taxable spending);
The estimate of sales tax collected from a CPS household is the household's spendable income times the sales tax rate. This relationship can be expressed as:
The final step in the estimation process is to allocate the estimated household sales tax contribution to the individuals in the household. We use total individual income as the basis for this allocation. Every individual with income receives credit for a proportionate share of the household's sales tax. Sales Tax Estimates. In fiscal year 199293, New York collected $6.28 billion in state general sales tax and $5.48 billion in local general sales tax for a total of $11.76 billion; for 199394, the totals increased slightly to $6.36 billion for the state, $5.71 billion for local, and $12.08 billion for total general sales tax collections (U.S. Bureau of the Census 1997).
Some of the sales tax collected does not come from the state's population. Some comes
from businesses and some from out-of-state residents. For the business collections, it is
appropriate to allocate those collections back to individuals since, ultimately, all business income
comes from individuals. However, for sales tax from spending by out-of-state residents, it is
difficult to rationalize how to distribute it to the various immigrant status groups within the state.
Thus, before allocating general sales tax collections to individual residents (and then to
immigrant status groups), we discount the total sales tax collections for spending by two
groups— out-of-state residents who earn income in New York and other nonresidents, such as
tourists. Full-year residents of New York report an average of 10.5 percent of the state's adjusted
gross income on New York State tax returns for tax years 1993 and 1994 (New York State
Department of Taxation and Finance 1996 and 1997). In addition to discounting the total sales
tax collections by this percentage, we allow another 10 percent for other nonresident sales.
Multiplying the total state sales tax collections of $12.1 billion by the complement of the
discounting factors gives a total of $9.6 billion for sales tax attributable to the resident population
(as represented in the CPS). To arrive at this total, we set
Differences between New York City and the balance of the state (Detailed Tables 1 through 6) represent the interplay of two factors. New York City has a higher tax rate than the rest of the state (Table X), but household incomes are lower in the city. Immigrants in New York City, in particular, have lower household incomes, especially when the refugees and undocumented immigrants are taken into account. Overall, however, the differences across nativity groups are very small— natives spend 3.0 percent of their income on sales tax versus 2.9 percent for legally present immigrants. Social Security The payroll tax commonly known as Social Security tax, also referred to as FICA, includes the old age, survivor, and disability portion (OASDI) and the health insurance (HI) portion.(38) Half the tax is deducted from employees' paychecks and the other half is paid by employers; self-employed persons pay both portions. For the estimates shown here, both the employee and employer shares are included in the aggregate estimates of taxes paid, but only the employee and self-employed portions are included when calculating the percent of income paid in taxes. TRIM2 simulates the Social Security taxes by directly calculating the amount owed on different types of earnings after first determining whether the CPS respondent is employed in covered employment using information on occupation and industry. The OASDI rate for 1994 was 6.2 percent each for employee and employer of qualified earnings below $60,600; the HI rate was 1.45 percent each for all qualified earnings. Rates for self-employed are double these rates. For agricultural and household employment, coverage ratios are used to adjust the TRIM2 estimates of taxes collected (Giannarelli 1992). The TRIM2 estimates of Social Security taxes are shown in Table 9. We estimate that $35.1 billion in Social Security taxes was paid in 1994 on behalf of 8.9 million employees in New York State and 0.6 million self-employed persons. For comparison, the table shows administrative data from 1992 and 1993,(39) even though the data are not strictly comparable; between 1992 and 1993, the administrative data changed from an employer basis to an employee basis, so time series data are inconsistent. The employee-based administrative data are consistent with the simulated number of workers— 9.4 million versus 9.5 million in the simulation, a difference of only 70,000, or 1 percent. The tax collections shown in the administrative data, $30.9 billion, are considerably less than the $35.1 billion from the TRIM2 simulation. However, the employer-based data show much larger tax collections one year earlier, $40.9 billion. The large differences could be attributed to the time difference (that is, collections for 1994 could have grown), or there still could be significant definitional differences between the data systems. For our estimates, we employ the TRIM2 estimates with no further adjustment.
All of the legally present immigrant groups pay a higher percentage of their income in Social Security tax (6.16.6 percent(40)) than do natives (5.7 percent). (See Table Z.) This tax is extremely regressive, which accounts for the lower proportion of income being paid by natives. The Social Security tax rate is 7.65 percent on wage and salary income below $60,600, 1.45 percent on wage and salary income above this amount, and zero on unearned income, such as interest, dividends, and capital gains. Unemployment Insurance Unemployment insurance is paid by employers on behalf of their employees. Thus, we include the unemployment insurance payments in our aggregate estimates, but do not include them as being paid from the employees' income. The insurance rate is 4.41 percent on the first $7,000 of the employee's income. TRIM2 simulates unemployment insurance directly, with the rate and base being determined by each CPS person's occupation, industry, and employment. According to our initial TRIM2 estimates, there are 8.5 million employees covered by unemployment insurance, substantially more than the actual number covered, 7.6 million (Social Security Administration 1997). Similarly, TRIM2 overestimates payments into the unemployment insurance system by 36 percent, $2.8 billion versus $2.1 billion. To bring the two figures into agreement, we multiply all estimated unemployment insurance payments by 0.75. Our final estimate of $2.1 billion differs from the administrative totals by only 45 million, or 2 percent. VII. ReferencesAhmed, Bashir and J. Gregory Robinson. 1994. "Estimates of Emigration of the Foreign-born Population: 1980-1990." Population Technical Working Paper No. 9. Washington, D.C.: U.S. Bureau of the Census, December. Clark, Rebecca L., Jeffrey S. Passel, Wendy N. Zimmermann, and Michael E. Fix 1994. "Fiscal Impacts of Undocumented Aliens: Selected Estimates for Seven States." Report to the Office of Management and Budget and the Department of Justice, Grant No. 94-IJ-CS-007. Washington, D.C.: Urban Institute. Department of Finance, Office of Tax Policy, City of New York. 1997a. "NYC Personal Income Tax Returns— Various Components by NYAGI Range: Tax Year 1994. Data provided by Michael Hyman, Director of Research, Tax Policy Division, 20 March. Department of Finance, Office of Tax Policy, City of New York. 1997b. "Taxable Assessment Profile, Citywide, FY 1997." Provided by Michael Hyman, Director of Research, Tax Policy Division, 30 April. Department of Taxation and Finance, New York State. 1997. New York Adjusted Gross Income and Tax Liability: Analysis of 1994 State personal income tax returns by place of residence. Albany: Office of Tax Policy Analysis. April. Department of Taxation and Finance, New York State. 1997b. Enactment and Effective Dates of Sales and Use Tax Rates. Publication 718A. March. Department of Taxation and Finance, New York State. 1996. New York Adjusted Gross Income and Tax Liability: Analysis of 1993 State personal income tax returns by place of residence. Albany: Office of Tax Policy Analysis. April. Espenshade, Thomas J. 1997. Keys to Successful Immigration: Implications of the New Jersey Experience. Washington D.C.: Urban Institute Press. Fix, Michael and Jeffrey S. Passel. 1994. Immigration and Immigrants: Setting the Record Straight. Washington, D.C.: Urban Institute. Garvey, Deborah L. and Thomas J. Espenshade. 1997. "State and Local Fiscal Impacts of New Jersey's Immigrant and Native Households." In Espenshade (ed.), Keys to Successful Immigration: Implications of the New Jersey Experience. Washington D.C.: Urban Institute Press. Giannarelli, Linda. 1992. An Analyst's Guide to TRIM2—The Transfer Income Model, Version 2. Washington D.C.: Urban Institute Press. Huddle, Donald. 1993. "The Costs of Immigration." Washington, D.C.: Carrying Capacity Network. Revised, July. Immigration and Naturalization Service. 1997. "Naturalizations." Chapter V. In 1996 Statistical Yearbook of the Immigration and Naturalization Service. Washington, D.C.: U.S. Government Printing Office. Immigration and Naturalization Service. Various years. Statistical Yearbook of the Immigration and Naturalization Service. Washington, D.C.: U.S. Government Printing Office. Internal Revenue Service. 1997. SOI Bulletin, Volume 17 (2, Fall). Washington, D.C.: U.S. Government Printing Office. Internal Services Division, Los Angeles County. 1992. Impact of Undocumented Persons and Other Immigrants on Costs, Revenues and Services in Los Angeles County: A Report Prepared for Los Angeles County Board of Supervisors. Los Angeles County. Lee, Ronald D. and Timothy Miller. 1998. "The Current Fiscal Impact of Immigrants and Their Descendants: Beyond the Immigrant Household." In Smith, James P. and Barry Edmonston (editors), The Immigration Debate: Background Papers on the Economic, Demographic, and Fiscal Effects of Immigration. Washington, D.C.: National Academy Press, in press. Massey, Douglas et al. 1994. "Theories of International Migration: The North American Case." Population and Development Review 19 (3): 431465. North, David S. and Marion F. Houstoun. 1976. The Characteristics and Role of Illegal Aliens in the U.S. Labor Market. Washington, D.C.: Linton and Co. Office of Refugee Resettlement. Various years, Refugee Resettlement Program: Annual Report. Washington, D.C.: U.S. Department of Health and Human Services. Office of Real Property Services, New York State. 1997. "Report of Effective Full Value and Tax Levy by Property Class Within School Districts for the 1994 Assessment Rolls at a 1993 Market Value Standard." Provided by Bill Heidelmark, 15 May. Passel, Jeffrey S. 1996. "Problem with March 1994 and 1995 CPS Weighting." Memorandum for CPS Users. Washington, D.C.: Urban Institute. November 12. Passel, Jeffrey S. and Rebecca L. Clark. 1997. "How Many Naturalized Citizens Are There? An Assessment of Data Quality in the Decennial Census and CPS." Paper presented at the annual meeting of the Population Association of America, Washington, D.C. March 1997. Passel, Jeffrey S. and Rebecca L. Clark. 1996. Taxes Paid by Immigrants in Illinois. Technical report to the Latino Institute of Chicago. Washington, D.C.: Urban Institute, May. Passel, Jeffrey S., Rebecca L. Clark, and Michael Fix. 1997. "Naturalization and Other Current Issues in U.S. Immigration: Intersections of Data and Policy." In Proceedings of the Social Statistics Section of the American Statistical Association: 1997. Alexandria, Virginia: American Statistical Association. Passel, Jeffrey S. and Joan R. Kahn. 1998. "Immigration, Fertility, and the Future American Work Force." Report to the Social Security Administration. Washington, D.C.: Urban Institute. Passel, Jeffrey S. and Karen A. Woodrow. 1984. "Geographic Distribution of Undocumented Immigrants: Estimates of Undocumented Aliens Counted in the 1980 Census by State." International Migration Review 18 (3, Fall 1984): 642671. Rea, Louis M. and Richard A. Parker. 1992. Illegal Immigration in San Diego County: An Analysis of Costs and Revenues. Report to the California State Senate Special Committee on Border Issues. San Diego: Rea & Parker, Inc. Rothman, Eric S. and Thomas J. Espenshade. 1992. "Fiscal Impacts of Immigration to the United States." Population Index 58 (3): 381415. Smith, James P. and Barry Edmonston. 1997. The New Americans: Economic, Demographic, and Fiscal Effects of Immigration. Washington, D.C.: National Academy Press. Social Security Administration. 1997. Social Security Bulletin: Annual Statistical Supplement, 1997. Washington, D.C.: U.S. Government Printing Office. U.S. Bureau of Labor Statistics. 1997. "Consumer Expenditure Survey: Tabulation." Website http://www.bls.gov/csxhome.htm. U.S. Bureau of the Census, Governments Division. 1997. "State Tax Collection Survey." Website http://www.census.gov/pub/govs/statetax. U.S. Bureau of the Census. 1995. Life Tables for U.S. Population Projections. Downloaded from the Population Division home page on the Internet, http://www.census.gov. Waite, Preston J. 1996. "CPS Supplements: 1996 March Weighting Specifications." Memorandum for Barry G. Fink. Washington, D.C.: U.S. Bureau of the Census, Demographic Statistical Methods Division. October 18. Warren, Robert. 1997. "Estimates of the Undocumented Immigrant Population Residing in the United States: October 1996." Washington, D.C.: Immigration and Naturalization Service, Office of Policy and Planning. August. Warren, Robert. 1994. "Estimates of the Unauthorized Immigration Population Residing in the United States, By Country of Origin and State of Residence: October 1992." Washington, D.C.: U.S. Immigration and Naturalization Service. Warren, Robert and Jeffrey S. Passel. 1987. "A Count of the Uncountable: Estimates of Undocumented Aliens Counted in the 1980 United States Census." Demography 24 (3, August): 375393. Word, David L. 1995. Unpublished estimates of nonimmigrants in the United States in 1990. Personal communication. Washington, D.C.: U.S. Bureau of the Census. VIII. Analytic Tables
Table 1. Population, by Nativity for States: March 1996 CPS
Table 2. Estimated Foreign-Born Population, by Legal Status and Country of Birth,
New York: 1995
Table 3. Undocumented Alien Population, by Country of Birth and State: October 1996
Table 4. Legal Status of the Foreign-Born Population by State: 1995-1996
Table 5. Federal Tax Returns for New York State, Tax Year 1994: Estimated by TRIM2
Table 6. New York State Tax Returns and Liability, Tax Year 1994: Estimated by TRIM2 and Administrative Statistics
Table 7. New York City Individual Tax Returns and Liability, Tax Year 1994:
Table 8. Residential Property Tax, by Nativity, Immigrant Status,
Table 9. Social Security Tax and Unemployment Insurance, New York: 1994 (Populations in thousands; taxes in millions of dollars)
Table 10. Taxes Paid by Type of Tax and Nativity/Legal Status, New York: 1995
IX. Detailed Tables
Detailed Table 1. Taxes Paid by New York Residents for Tax Year 1994,
Detailed Table 2. Percent of Tax Paid for Tax Year 1994 for Nativity Groups by Type of Tax,
Detailed Table 3. Per Capita Income and Taxes Paid by New York Residents for Tax Year 1994, by Nativity/Immigration Status, Type of Tax, and Area
Detailed Table 4. Taxes Paid by New York Households for Tax Year 1994,
Detailed Table 5. Percent of Tax Paid for Tax Year 1994 for Households
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Outside New York City, at the
individual level, the legal foreign-born, on average,
contribute $8,400 in total taxes, compared with
$6,800 for natives. Tax contributions for
naturalized citizens ($10,700) are particularly
high. The tax contributions reflect income
differences. The average total individual
income of the legal foreign-born ($23,900)
surpasses that of natives ($19,100). Again, per
capita incomes for naturalized citizens
($30,000) are particularly high. (See 
These findings suggest that, in cost-benefit
analyses of the impact of immigrants, if costs of providing
services to the U.S.-born children of immigrants are
attributed to immigrants, then the tax contributions of
adults whose parents were immigrants should be attributed to immigrants as well.
Where Do Taxes Paid by the Legal Foreign-Born Go, to the Federal Government or to
State and Local Governments? How Does This Compare with Natives?
Refugees and Asylees. In any given year,
refugees tend to come from only a relatively small
number of countries. In addition, countries that
send refugees usually do not send large numbers
of other legal immigrants in the same year. For
each two-year period
= percentage of illegal aliens, i, aged 1864 of sex s in New York in occupation x among aliens who applied for legalization under Section 245A of IRCA (from LPS);
= number of illegal aliens in New York in March 1995 (Warren 1997); and
= percentage of illegal aliens, i, in New York who are aged 1864, of sex s (from LPS).
if they meet any of the following criteria:
The weighting adjustment proceeds sequentially.
First, to derive final weights for refugees, the CPS weights
for refugees are multiplied by the ratio of the estimated
population to the CPS population for each period of entry
for each of two country groups shown in the second
column of 
How the incomes of legally present immigrants stack up against natives depends on the
measure used. At the individual level, incomes of the legal foreign-born are about equal to those
of natives, $18,000 versus $18,100. (See 
Among the legally present foreign-born,
naturalized citizens and nonimmigrants have the
highest incomes, LPR aliens have somewhat
lower incomes, and refugees have substantially
lower incomes. (See 
For LPR aliens, naturalized citizens, and refugees,
per capita incomes go up as time in the United States
increases. (See 
Incomes and tax contributions of second-generation Americans— U.S. natives with one
or two parents who are immigrants— are remarkably similar to those of third-and-higher-
generation Americans. (See 
The high income group pays a very high
percentage of federal income taxes because federal
income tax is extremely progressive; that is, as
income increases, the tax rate goes up. For example,
in husband-wife households, income over $250,000 is
taxed at a rate of 39.6 percent, while income between
$38,000 and $91,850 is taxed at a rate of only
28.0 percent. As a result of the rate structure,
households with incomes over $200,000 pay
34 percent of their income in federal income tax. The
98.8 percent of households with incomes below
$200,000 pay only 11 percent of their income in
federal income tax.
New York State Income Tax Estimates.
Natives pay a higher percentage of New York
State income tax, 84.8 percent, than their
representation in the population, 81.8 percent
(
New York City income tax is somewhat less
progressive than the state income tax, but natives still pay a higher percentage of the tax
(71.1 percent) than their representation in the population (65.8 percent). (See 
; 
; 
;
; and
.
and 
, respectively) are calculated as follows:
. The same
technique is used to estimate residential property taxes of native heads and their spouses in rental
units, and of foreign-born heads and their spouses in owned and rental units.
, we assume
that individual property tax contributions are proportionate to individual rental income:
), renters (
), and landlords (
) are, therefore:
. The same technique is
used to estimate residential property tax attributable to renters.
, we use the
same technique used for the New York City estimates:
.
Although, on average, the legal
foreign-born pay about as much in residential property
taxes as natives, the means through which they
pay this tax differs substantially. On average, the
residential property tax the legal foreign-born
pay on the dwellings they inhabit, either as
owners or renters, is slightly lower than the
amount paid by natives, while the residential
property tax they pay as landlords is substantially
higher than that of natives.
This large contribution of long-term naturalized
citizens is due to the fact that this group is far more likely to
be landlords than any other, 12.4 percent, compared with
3.7 percent for natives, and 3.8 percent for the state population
overall (
= 1, if a 2-unit dwelling; = 0, otherwise;
= 1, if a 3-9 unit dwelling; = 0, otherwise;
= 1, if a 10+ unit dwelling; = 0, otherwise;
= total household income (in thousands of dollars);
= age of householder; and
= number of persons in household.
= 1, if householder is 65 years or older; = 0, otherwise;
= number of persons with earnings in household; and
, is estimated as food costs per person times the
household's size:
= multiplicative scaling factor (ultimately set to 0.60);
= total household income;
= percent of income spent on housing (from the housing equation);
= household spending on food (from the food equation); and
= number of adult immigrants aged 18 and over with at least $5,000 in
earnings.
In New York, the sales tax has a state component that
is uniform throughout the state and county and city
components that vary across jurisdictions. In the March 1995
CPS, 11 metropolitan statistical areas (MSAs) plus New York
City can be identified. For each of these areas, we calculate
the state plus local sales tax rate as a population-weighted
average of the tax rates for the counties and cities that make
up each MSA and the balance of the state. These average
sales tax rates are shown in 
in the spending equation to 0.60.
Social Security Tax Estimates. Social
Security tax is the second largest of the taxes we
estimate; the $34.7 billion collected from this
tax is less than the $45.8 billion collected in
federal income tax, but far greater than the
$13.5 billion collected in state income tax. For
Social Security tax, legally present immigrants,
who pay $5.7 billion, actually pay a higher
percentage of the tax, 16.7 percent, than their
representation in the population, 15.7 percent.
However, naturalized citizens are the only
immigrant group that pays more than their
proportion in the population.