Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 297
--> 7 The Future Fiscal Impacts of Current Immigrants Introduction Chapter 6 is a snapshot of the effects immigrants have on the U.S. fiscal situation, on an existing group of new immigrants at one point in time. It assigns to native-born residents alive at that time the tax burdens and expenditure benefits occasioned at all levels of government by members of different immigrant cohorts at that time. That picture is instructive, but it cannot be used to predict the long-term consequences of current or new immigration policies. This chapter sets out a forward-looking projection of the long-run implications immigration has for the fiscal balance. It fills in the picture drawn by the static calculations. For our purposes, those static calculations in Chapter 6 have important limitations. These are particularly relevant when we consider the impact of immigrant flows that change in size and character over time, and when we wish to determine how such changes in the immigrant pool will affect fiscal balance. The limitations in the static calculations stem from four factors. First, because a static calculation takes one group of existing immigrants, it combines members of different immigrant generations in that group who differ from one another. It therefore gives an inaccurate picture of the impact of any particular generation of immigrants. If, for example, elderly immigrants generally have had a higher level of income than younger, more recent immigrants, then the ratio of their Social Security benefits (which are based on their past incomes) to the payroll taxes of the new immigrants (based on their own, lower incomes) will overstate the ratio based on the benefits and taxes of either group separately. Second, even if immigrants are all the same, tax and expenditure rules may
OCR for page 298
--> change in the future. Today's Social Security benefits may be more generous than benefits in the future, so we cannot use current benefits to infer the burden that today's immigrants will ultimately impose on the system when they age. Third, the outcome of the static calculation depends both on the relation of taxes to expenditures at each age over the life cycle and on the relative numbers of immigrants of each age in the population. Because patterns of expenditures and taxes differ over a lifetime, no single snapshot can accurately depict the impact of any individual cohort of immigrants if the population's age structure changes over time. For example, education spending relates primarily to the school-age population, and income tax payments relate to the working-age population. Consequently, increasing cohort size will overstate the cost of educational expenditures relative to the revenue from income taxes when these children enter the labor market. However, because any single cohort receives its education benefits earlier than it pays income taxes, the failure to discount income taxes overstates revenues relative to education spending. In general, unless the rate of growth of populations plus productivity equals the appropriate government discount rate, these effects will not cancel one another. Fourth, because the government's budget need not be balanced over any particular time period, a deficit (or surplus) will develop equal to the difference between revenues and expenditures. There is no obvious way to assign the incidence of a deficit in a static calculation; if expenditures exceed revenues, this is not an "error" to be corrected. Yet we know that running a budget deficit today alters the fiscal policy choices tomorrow. Many static calculations suffer from an additional, serious problem if they are based on the analysis of households with immigrant heads. In this case, they miss the effect of the adult native-born children of immigrants who do not live in households headed by immigrants, as we mentioned in Chapter 6. Because these younger people are likely to be making substantial contributions to the fiscal balance, their omission biases the results toward negative outcomes. This difficulty is not intrinsic to the approach, however. These problems highlight the inherent limitations of static calculations, but they also point to a solution. Using the static as a starting point, we can project revenues and expenditures into the future, taking account of differences in individuals, policies, cohort sizes, and budget deficits to arrive at a more meaningful calculation that assigns revenues and expenditures to groups of immigrants at each date. This approach will yield the net impact of each group, based on the present value of these annual flows. Dynamic Incidence The methodology of dynamic calculations of incidence is, to a large extent, simply the methodology of static calculations. Initially, we must go through the same exercise of estimating the marginal tax payments and benefits by age of
OCR for page 299
--> different groups, determining the extent to which the addition of an individual would change expenditures or revenues, holding fixed the fiscal position of those already in the population. Once this is done for the current period, though, we must take several additional steps and make many more assumptions to complete the calculation. These additional steps include projecting future taxes and expenditures, discounting these future flows, and defining the nature of an immigrant "experiment." Projecting Future Taxes and Expenditures Because the dynamic calculation is forward-looking, it requires estimating the trajectory of taxes and expenditures far into the future. These estimates affect the calculation in potentially important ways, since we are calculating the present value of the difference between taxes and expenditures at each age that immigrants pass through. If the government runs a large deficit in future years, then expenditures will exceed taxes for everyone on average, including immigrants. If taxes are raised in the future to balance the budget, or if expenditures are reduced, the negative impact of immigrants will decline, or the impact will become more positive. The rules governing taxes and expenditures change every year, and it is impossible to predict precisely how they will evolve over the relevant future. Over the long run, however, any government faces an overall constraint on its ability to use deficit finance, which narrows the range of possible outcomes. In particular, it cannot let its debt grow without limit relative to the economy, as measured by gross domestic product (GDP), without losing credibility in its ability to repay and may eventually face default. To reflect this, it is necessary to assume that the ratio of debt to GDP stabilizes at some point. From this assumption comes the overall changes in taxes net of expenditures necessary in each year, relative to current policy. To see why it is necessary to make some assumptions about a future fiscal adjustment, consider for a moment what would happen if we simply let taxes and expenditures follow a pattern that adheres to current rules. We initially projected taxes and expenditures each year, assuming a particular rate of economic growth, a particular pattern of immigration, and maintenance of current fiscal rules (except those already slated to change, like Social Security provisions). The results of that projection for the pattern of national debt is displayed in Figure 7.1. This figure depicts the time path of the national debt that will emerge given the present U.S. fiscal picture. Current tax and expenditure policies will cause the debt to explode over time (Auerbach, 1994; Congressional Budget Office, 1996). We can consider a variety of changes in taxes and expenditures that will bring the path of national debt into line with the particular assumption about how it will be stabilized, and measure the incidence for each scenario. For example, we might assume that the debt/GDP ratio is stabilized immediately at its current value; that
OCR for page 300
--> Figure 7.1 Gross federal debt as percentage of gross domestic product. (Historical data taken from 1996 Statistical Abstract, Table 512.) Uncontrolled: no fiscal controls on deficit. Adjusted: adjusted to historical data series using 1990 as the base year. current policy remains in place for 10 years, after which the ratio is stabilized; or that current policy remains in place until the debt/GDP ratio hits 1.0. For each of these scenarios, we can consider the impact of adjustment in income taxes, in transfer payments, in defense spending, or in any combination of these and other components of the budget. This approach to the government's long-run budget constraint clarifies the appropriate treatment of government debt and deficits under a dynamic incidence calculation. The burden of the debt itself is not assigned directly. Rather, individuals and their descendants are assigned the higher future taxes or lower future benefits that a higher current deficit may necessitate. In each instance, we perform these calculations as a "partial-equilibrium" exercise. That is, we estimate budget changes needed under the assumption that people do not change their behavior in response to the new conditions. This assumption is unrealistic, but it is necessary given the complexity of the calculation. In any event, this limitation is one carried over from the static-incidence approach, and it is not likely to alter the qualitative nature of the conclusions. One general equilibrium issue, however, needs to be discussed—how the net increase in the incomes of natives discussed in Chapters 4 and 5 would affect the
OCR for page 301
--> fiscal calculations. In general, one can think of the total effect of immigration on native after-tax income (in a particular year) as the sum of the net increase in income from labor market effects plus the net fiscal impact in that year. Thinking of the labor and fiscal effects as completely independent seems problematic at first glance, because the increase in income from the labor market effects would presumably lead to additional tax payments, which one would think ought to be considered in doing the fiscal calculation. However, in adding these two pieces together, it is important to realize that it would be double counting to first count a gain in income in the labor market effects and then also count the part of that increase that goes to additional tax payments in calculating fiscal effects. How should one factor in the increase in incomes of natives in doing the fiscal calculation? The answer depends on what is assumed about the path of future taxes and expenditures. Consider the case in which taxes and benefits adjust to stabilize the debt/GDP ratio at some point. For simplicity, think of benefits as fixed, so debt targets are met through adjustments in taxes. The total amount of tax revenues needed to meet the debt target in a given year is essentially unaffected by the increase in the incomes of natives, although the tax rate needed to generate that level of revenues will fall. Taking into account the income gain to natives may thus lead to a slight shift in the incidence of the tax that would be unfavorable to natives. With a reduced tax rate for everyone and higher income for natives, natives would pay a slightly higher portion of total tax revenues, but this effect would have to be small. Aside from this shift in incidence, it would be correct to simply ignore the additional tax revenues coming from the gain in native income in the fiscal calculations, because it is rebated to natives through lower tax rates. Discounting Future Dollar Flows To discount the future flows for each immigrant or immigrant group, we must also settle on an appropriate discount rate. Here, there are a variety of options. The straightforward approach is to use a government borrowing rate, which will provide, in expected value, the present value of future net flows. However, given the uncertainty of the future and the riskiness of future taxes and benefits, it may be more appropriate to discount these flows with a discount rate that reflects their risk characteristics. For example, future income taxes might be discounted with a market discount rate that reflects the riskiness of future income (see Auerbach et al., 1991, for further discussion). Because the "right" discount rate depends on the question being asked, calculations based on a range of discount rates may be appropriate, and that is the approach we follow. Defining the New Immigration "Experiment" A dynamic-incidence calculation is intended to enable us to determine the
OCR for page 302
--> fiscal impact of an additional immigrant of a particular type at a particular date. But we must be more precise regarding this "experiment." An immigrant's arrival has fiscal consequences not only from the immigrant directly, but also from her offspring and their descendants, even though they themselves will be native-born. Thus, we must include in the calculation changes in taxes and expenditures associated not only with the immigrant, but also with her descendants. This process relies on the assumptions made about the characteristics of future immigrants and the speed of assimilation. To calculate the future tax and expenditure flows for immigrants, we must estimate the characteristics of new immigrants, as well as the extent to which the differences between immigrants and natives (in, for example, the birth rate, earnings conditional on education, the fraction of those who are eligible for a benefit that actually apply for it) disappear over time through assimilation. It may also depend on assumptions about the extent to which immigrants marry outside their own ethnic groups, to the extent that this is deemed to influence the rate of assimilation. Indeed, the dynamic-incidence approach should also allow us to compare the impact of a new immigrant of a particular type at a particular date to that of a comparable native birth. This comparison is useful in separating the fiscal impact of immigration into the impact of population growth generally and the impact of growth through immigration. Kinds of Impacts The lifetime fiscal effects of an immigrant and his descendants can be divided into two categories: first, the fiscal benefits or costs of adding one more person to the population regardless of immigrant status and, second, the fiscal benefits and costs associated with the special characteristics of immigrants, such as age at arrival, time since arrival, English language ability, and education. We will briefly discuss these two categories of impacts. Fiscal Impacts Relatively Independent of Immigrant Characteristics Any increment to the population, holding all else equal, will have fiscal effects. These arise in part because a larger population helps to bear the cost of so-called public goods—those that provide services to all in the population at a cost that does not rise with the size of the population. National defense, expenditures on veterans, and research on health and science all are public goods.1 The cost per capita of providing a given level of services declines as population rises because more taxpayers share the unchanging total costs. Also, a larger popula- 1 For a discussion of empirical estimates of the "publicness" of various kinds of government expenditures, see Chapter 6.
OCR for page 303
--> tion helps to bear the burden of the preexisting public debt through tax payments to cover interest or repayment charges.2 Much like anyone else in the population, immigrants use services that are costly to provide, or that others can use less freely—so-called congestion costs. Examples include services from roads, sewers, police and fire departments, libraries, airports, and foreign embassies. These services may have a public good aspect, but because they are highly congestible, we treat them as if immigrants raise both the demand for them and the cost of meeting that demand, in proportion to their numbers. Such items have both a capital cost and a current cost, and we account for these separately. Additional members of the population, whether immigrant or not, crowd the existing social infrastructure, including roads, libraries, airports, sewage and water supply systems, and public buildings. We include in our analysis a cost of investment for each incremental immigrant to replicate the existing social capital stock. We have done this in two ways. First, when we include the present value of a per capita share of government expenditures on congestible goods and services, we implicitly include this capital cost because capital outlays are an item in such expenditures. Second, we use a direct estimate of the per capita value of net public capital (U.S. Department of Commerce, 1994), and use an annual flow of services plus depreciation of this capital, while omitting the capital outlays from expenditures on congestibles.3 These two methods yield nearly identical results. Sensitivity of Fiscal Impacts to Immigrant Characteristics Apart from their simple numbers, the specific characteristics of immigrants influence their fiscal impacts. Immigrants arrive with human resources different from those of the rest of the population. In recent years, for instance, there has been concern that immigrants are disproportionately poor and uneducated. Such immigrants may both pay less in taxes and receive more in benefits than natives do. An immigrant's age at arrival is also important. Natives begin life in the United States when they are born, but an immigrant can arrive at any age; the modal age is around 25. If she arrives after school age, although she may have less education than natives, the public costs of her education have been borne by the sending country. Although she may give birth to children in the United States 2 Just as a larger population helps dilute the costs of past obligations, it also dilutes the per capita value of publicly owned wealth of various kinds (national parks and forests and publicly owned mineral rights, for example). The calculations below do not include such effects. 3 The estimate of total net public capital is taken from Survey of Current Business (1994). The annual flow is obtained by multiplying the per capita value, $17,000, by the assumed rate of interest plus a rate of depreciation of 4 percent. The descendants of the immigrant are also taken into account, and a net present value is then calculated.
OCR for page 304
--> who will then incur educational costs, these costs will be paid in part by the immigrant parents, who will typically be working and paying taxes. A child born here or a young immigrant child, however, will not generate incremental tax payments until he or she begins to work, since the parents would have been paying taxes in any case. At the other end of the age spectrum, elderly immigrants, arriving late in their working years or during retirement, will be particularly expensive, since they can qualify for certain kinds of benefits such as Medicaid and Supplemental Security Income, even if they do not qualify for Social Security and Medicare. Consequently, the age distribution of arriving immigrants distinguishes them from native increments to the population, all of whom ''arrive" at age 0. The different fertility, mortality, and emigration rates of immigrants also have effects. On one hand, on average, immigrants have higher fertility and lower mortality rates than natives do, which affect the benefits they receive and the fiscal impact of their descendants. On the other hand, a substantial proportion of immigrants (about 30 percent) return to their country of origin, presumably taking at least their younger children with them, thus substantially mitigating the effect of their higher fertility.4 The need for bilingual education for many immigrants makes their public education more costly than that of natives. Cultural factors may influence the extent to which immigrants make use of the benefits for which they qualify. For example, strong family values may reduce the use of nursing homes by elderly immigrants. Finally, immigrants and their descendants are concentrated in certain areas of the country, and these areas may have different taxes and benefits from the rest of the country. For example, the states in which immigrants concentrate on average have higher levels of per pupil expenditures in public schools, and also higher state and local taxes. To put these various fiscal impacts into perspective, it is useful to consider how important the programs involved are in current federal expenditures. In 1995, for example, expenditures on what we have categorized as public goods accounted for 23.7 percent of total federal outlays. Age-related expenditures made up an additional 55.4 percent of federal outlays, with debt payments accounting for 13.9 percent, and spending on congestible goods or social infrastructure making up the last 7.0 percent. Immigrants affect each of these categories differently in our calculations. New immigrants reduce the burden to natives of public goods and interest on the public debt; on average they are younger than 4 As noted in Chapter 3, there is little evidence on the characteristics of return migrants. However, there seems good reason to think that they leave few children behind them. Most return migrants leave the United States within the first decade or so of their arrival. Emigration thus cuts short the period during which they could have children in the United States. This also means that children born in the United States are likely to be young when their parents emigrated, making it unlikely that they would leave them behind.
OCR for page 305
--> natives when they arrive, so immigrants initially participate less in age-related entitlements. The annual or cross-sectional estimates of fiscal impacts were done for state and local governments as well as at the federal level. However, it does not make sense to do the longitudinal estimates for individual states or localities, because there is so much mobility from locality to locality and from state to state. Each year, 17 percent of the U.S. population changes residence, 6 percent changes county, and 3 percent changes state. For this reason, it does not make sense to do calculations that are based on the assumption that people remain in the same state over their lifetimes, and that their descendants do the same. When we do these calculations at the national level, we can simply group together all the state and local expenditures. One drawback, however, is that we can easily lose sight of the fact that immigrants and their state and local fiscal impacts are very heavily concentrated in a few states, rather than evenly spread across the nation. Recall from the introduction that we do not take into account indirect fiscal effects of immigrants arising from any consequences of immigration for the earnings or employment of the existing labor force. This means that we will not consider the possibility that immigrants impose fiscal costs indirectly, by causing native workers to become unemployed or to drop into poverty due to reduced wages. The earlier chapters on the labor market effects of immigration suggested that any such negative effects on native workers are likely to be quite small, and the effects could even be positive. However, the possibility remains that immigration into a particular state may cause some out-migration of workers to other states, resulting in fiscal effects for high-immigration states that we have not taken into account. For the nation as a whole, such effects should average out to zero. A Word on the Demographic Unit of Account Many studies of the fiscal impact of immigrants use the immigrant-headed household as the unit of account. Here we use the individual as the unit of account. We do so because it is necessary for longitudinal calculations. If we were to use households, we would have to deal with changing household structure over time through marriage, divorce, widowhood, the departure of growing children, the arrival of additional family members from abroad, death of elderly members, and so on. We would also have to deal with nonimmigrant household members. Using the individual is much simpler in all these respects. Note that our calculations can subsequently be used as the basis of constituting families or households of immigrants when needed for interpreting the results, or for comparing them to results of studies that use the household framework.5 5 The age profiles reflect the average payments of taxes and receipt of benefits for all immigrants at the age in question. Thus attempts to reconstitute specific family configurations by combining the individual profiles at particular ages will also yield families that were implicitly assumed to pay
OCR for page 306
--> Some Data Issues Most of our analysis is based on the Current Population Surveys (CPS) of March 1994 and 1995. The relative merits of several data sets for this purpose—the Current Population Survey, the Survey of Income and Program Participation, the Public Use Micro Sample—are discussed in Appendix 7.C. The number of respondents in the combined CPS sample is roughly 300,000, of which about 29,000 are foreign-born. The 1994 and 1995 March CPS were simply pooled, treating each as a separate sample. Because of the way respondents are rotated from one panel to the next, approximately one-half of the data represent reinterviews of the "same" household a year later. In principle, these surveys cover illegal immigrants as well as legal immigrants and nonimmigrants (foreign students and foreign business travelers). To the extent that these are included in the CPS, they distort the information about immigrants, particularly those who don't stay very long. The problem may not be trivial, since the number of nonimmigrants in the United States at any time is comparable to the annual inflow of immigrants. At shorter durations in the United States, our calculations could most accurately be said to apply to the foreign-born, rather than to immigrants per se. In practice, we do not know the coverage of illegal immigrants, but we suspect that it is incomplete. In our analysis we cannot distinguish between legal and illegal immigrants. Presumably, illegal immigrants both pay less in taxes and receive less in benefits than other immigrants do.6 The Heterogeneity of Immigrants and Intra- and Intergenerational Mobility The fiscal impacts of immigrants vary greatly depending on a number of their characteristics. The benefits received by immigrants in the United States average taxes and receive average benefits for people of the ages in the family, and these amounts may be out of line with their exact circumstances. However, since both age and education level of self or parent could be taken into account, this does not seem to be a serious problem for constituting families. For example, low-education families would be more likely to receive Aid to Families with Dependent Children, and this would be reflected in the average numbers. 6 Readers should also note that the calculations in this chapter are based on estimated relationships between immigrant status, tax payments, benefit receipts, and the like. That is, we use data to calculate these relationships in the CPS sample, and then make inferences about immigrants and natives in general. Inevitably this introduces some margin of error, as a randomly chosen sample is very unlikely to have exactly the characteristics of the entire U.S. population, and there is some measurement error involved in collecting information through any survey. We have no reason to believe that these are misleading estimates, but estimating these relationships adds additional uncertainty about how close our projections of the effects of an additional immigrant would be to the actual effects.
OCR for page 307
--> and the taxes they pay depend strongly on their earnings. Migrants to the United States do not all have the same human resources, which heavily influence earnings. Our analysis captures some of this heterogeneity by distinguishing three categories of educational attainment of immigrants and others: less than high school, exactly high school, and more than high school. For our longitudinal analysis, we need to project educational attainment for the children and grandchildren of immigrants, according to the immigrant's education at arrival. For this purpose we analyzed data from the General Social Survey in a merged sample, calculating intergenerational transition matrices for these three educational categories. Our analysis distinguishes three lengths of an immigrant's time since arrival: less than 5 years, 5 to 9 years, and 10 years and more. For every age, for every program, and for taxes, we estimate three age schedules from the CPS data, one for each duration. That is, we fully incorporate all the ways in which age and duration of residence here interact for these three duration categories and five-year age groups. The durations were chosen to correspond to potential qualification for entitlement programs, such as Aid to Families with Dependent Children (AFDC) and Old-age, Survivors, Disability, and Health Insurance (OASDHI).7 We also analyze the taxes paid by immigrants according to their education, age, and time since arrival. Because immigrants arriving in different periods have had different characteristics, the time of arrival variable may overstate the degree of earnings progress that can be expected of current immigrants. Those who have been in the United States for more than 10 years in 1994-95 may be different from more recent arrivals in ways not fully captured by their education and age. For this reason we have done the analysis in two ways: first, assuming that the taxes paid by immigrants do follow the progress indicated by the trajectory estimated according to "time since arrival" and, second, assuming that the earnings follow this trajectory only for the first 10 years, and that thereafter the ratio of earnings between immigrants and native-born workers is fixed.8 There are 7 For immigrants who arrive after age 55, we constrain benefits after 9 years' duration to conform to those for 5 to 9 years' duration. This procedure avoids confusing qualifiers and nonqualifiers for OASDHI for people at age 70 and above at durations of 10+ years. However, many elderly immigrants whom the data describe as having been in the United States for only a short time, and who therefore would not be expected to qualify for many benefits such as OASDHI, must actually be earlier immigrants who worked in the United States long enough to qualify for OASDHI and other programs, and who have recently returned after a stay abroad. The duration variable will be misleading for such people, and this creates problems for estimates of the costs of elderly first-time immigrants, and for calculating the effects of the 1996 welfare reform legislation on the costs of elderly immigrants. In addition, some elderly first-time immigrants are refugees, who qualify for certain benefits despite their short durations of residence in the United States and lack of U.S. work histories. For these elderly immigrants, the duration variable should have different effects. 8 Duleep and Regets (1976) compare the 1994 and 1995 CPS data and conclude that the longitudinal changes accurately reflect the cross-sectional duration profiles.
OCR for page 352
--> congestible goods would double the positive NPV. Help from immigrants in paying the substantial costs of future population aging and rising health care costs for the elderly is also a major factor. Special characteristics of immigrants also act to raise the NPV: they tend to arrive in the early working years; some of their children are likely to be educationally upwardly mobile, and they receive lower benefits than others. By contrast, that immigrants have lower education than natives, and consequently pay less in taxes, reduces the NPVs. While the average long-term fiscal impacts of immigration are generally found to be positive, an increase in the annual flow of immigrants would, for a couple decades, have a negative fiscal impact overall. The timing and extent of such a period would depend on federal fiscal policy. At the state and local level, the annual fiscal impact would remain negative for many decades, and the overall NPV would be negative. At the federal level, the annual fiscal impact would be positive from the start. We have also calculated the annual fiscal impact per U.S. resident of an increase of 100,000 per year in the immigrant flow under the baseline assumptions (change the sign for a decrease). We found it to be roughly +$30 per person, composed of +$40 at the federal level and -$10 per person at the state and local level. This average amount per person strikes us as rather low, despite the large NPVs. However, it should be kept in mind that an earlier period of greater deficits is followed by a later period of greater returns. With due account taken of the many uncertainties in our estimates, it appears unlikely that immigrants and their descendants impose worrisome costs at the combined federal and state and local levels for the average U.S. resident. Indeed, our calculations suggest that immigrants may instead, on average and in the long run, have a positive fiscal impact. Nonetheless, immigrants with certain characteristics, such as the elderly and those with little education, may be quite costly. And residents of certain states with large shares of immigrants without doubt bear higher costs that in some cases may not be offset by the broadly shared gains at the federal level. Key Conclusions Households headed by immigrants include the native-born school-age children of immigrants, who incur high costs of public education. However, they do not include the working-age native-born children of immigrants, who typically have a positive fiscal impact. For this reason, cross-sectional or current fiscal impacts estimated for immigrant-headed households are biased toward negative numbers. The relative intensity of program use by immigrants and natives at the federal, state and local levels varies significantly from one program to another. For example, immigrants and their young children use bilingual education, SSI, public assistance, and Medicaid more heavily than others in the population, but they also use Social Security and Medicare more lightly than do others.
OCR for page 353
--> The implication of this diversity is that we must be comprehensive in our examination of program use to obtain an accurate assessment of the fiscal impacts of immigration. We find that the pattern of per capita overall program use at each age is very similar for immigrants and others, with immigrants and their children imposing somewhat higher costs at the young ages, and lower costs above age 50. The cost of benefits used by immigrants and their young children is actually 8 percent less per capita than the costs of benefits used by the rest of the population, in part because their age distributions differ. On average, immigrants pay considerably lower taxes at each age than do others, and overall they and their native-born children under the age of 20 pay about one-third less than does the rest of the population, again due in part because of their age distribution. In assessing the long-term fiscal impact of immigrants, it is important to take into account the likely descendants of the immigrants, and the likely educational attainment of these descendants. The long-term (net present value) fiscal impact of an immigrant varies greatly across different types of immigrants. Some groups of immigrants bring net fiscal benefits to natives, and others impose net fiscal costs. Among other things, these fiscal effects depend heavily on the characteristics of immigrants, including age at arrival in the United States, educational attainment, and time spent in the United States. The net present value typically peaks for ages at arrival of 10 to 25, and then declines to a trough for those arriving in their late sixties. This curve is higher, the higher the education. Under our baseline assumptions, the average fiscal impact (net present value) of an immigrant with less than a high school education is -$13,000, and that for an immigrant with more than a high school education is +$198,000. Similarly, older immigrants impose significant fiscal burdens, and younger immigrants produce fiscal surpluses. Averaging across these characteristics, immigrants under our baseline scenario have a negative fiscal impact at the state and local level, but a larger, positive impact at the federal level, resulting in an overall positive impact for the United States. Under most scenarios, the long-run fiscal impact is strongly positive at the federal level, but substantially negative at the state and local level. The federal impact is shared evenly across the population, but these negative state and local impacts are concentrated in the few states that receive most of the immigrants. The average fiscal impact of immigrants under the baseline assumptions is positive in part because they tend to arrive at young working ages, in part because their descendants are expected to have higher skills and incomes, in part because they pay taxes for some items, such as national defense and interest on the federal debt, for which they do not impose costs, and in part because they will help to pay the public costs of the aging baby-boom generations.
OCR for page 354
--> The 1996 Personal Responsibility and Work Opportunity Reconciliation Act, by prohibiting new immigrants from receiving means-tested benefits for a period after arrival, will make the long-term fiscal impact per immigrant more positive by about $8,000. If the long-term federal budget imbalance is not addressed—which is not a realistic policy option—the overall fiscal impact of immigrants will be slightly negative; if the debt/GDP ratio is held constant starting immediately, or starting in 2016, the overall fiscal impact will be positive. It will be more negative if discount rates are higher, and if budgetary imbalances are addressed by reducing benefits rather than by raising taxes. Although the average long-term fiscal impacts of immigration are generally found to be positive under most scenarios that we tried, the overall annual fiscal impact of an increase in the annual flow of immigrants would be negative for a couple of decades before it turned positive. The timing and extent of such a period would depend on federal fiscal policy. References Auerbach, A.J. 1994 The U.S. Fiscal Problem: Where We Are, How We Got Here, and Where We're Going. NBER working paper #4709. Cambridge, MA: National Bureau of Economic Research. Auerbach, A.J., J. Gokhale, and L.J. Kotlikoff 1991 Generational accounts: A meaningful alternative to deficit accounting. Pp. 55-110 in David Bradford (editor), Tax Policy and the Economy. Cambridge, MA: MIT Press. Board of Trustees of the Federal Old-Age and Survivors Insurance and Disability Insurance Trust Funds 1996 Annual Report of the Board of Trustees of the Federal Old-Age and Survivors Insurance and Disability Insurance Trust Funds. Washington: U.S. GPO. Clark, R.L. 1994 The Costs of Providing Public Assistance and Education to Immigrants. PRIP-UI-34, Program for Research on Immigration Policy, The Urban Institute. Congressional Budget Office of the United States 1996 The Economic and Budget Outlook: Fiscal Years 1997-2006 Available from U.S. Government Printing Office, Washington, D.C. Duleep, H.O. 1994 Social security and the emigration of immigrants. Pp. 97-128 in Migration: A Worldwide Challenge for Social Security. Geneva: International Social Security Association. Duleep, H.O., and M.C. Regets 1997 Measuring immigrant wage growth using matched CPS files. Demography 34(2):239-249. MaCurdy, T., T. Nechyba, and J. Bhattacharya 1996 An Economic Framework for Assessing the Fiscal Impacts of Immigration. Paper prepared for the Panel. Department of Economics, Stanford University, September. Sheffrin, S., and M. Dresch 1995 Estimating the Tax Burden in California. Berkeley, CA: California Policy Seminar. U.S. Department of Commerce 1994 Fixed Reproducible Tangible Wealth in the United States: Revised Estimates for 1991-1993 and Summary Estimates for 1925-1993. Survey of Current Business, Bureau of Economic Analysis.
OCR for page 355
--> Appendix 7.A Intergenerational Educational Mobility The General Social Survey (GSS) has been carried out every year since 1972 by the National Opinion Research Center (NORC). It collects information on the nativity of the respondent's parents and grandparents, on the respondent's education, and on the education of the respondent's parents. We categorized all educational attainments as less than high school, high school, or more than high school. For a specific generation of respondent (first, second, or third, based on the nativity questions), we created a set of all linked pairs of parent-child educational attainments, using the information on education of respondent and of each parent of the respondent. We then treated the parents as the reference unit, and for each parental level of education, we calculated the proportional distribution of the children by educational attainment. Even within parental education categories, the children of higher-fertility parents have lower educational attainments. However, we did not weight for this, since the overrepresentation of such children in the sample simply reflects the higher fertility of their parents, which we wish to be reflected in the results for the average parent. We ignore differences in fertility by education of parent, which could bias the results in the optimistic direction. It is well known that the characteristics of immigrants have been changing over the past several decades. Our controls for educational attainment may be insufficient to capture these changes. As a further control, we estimated separate transition matrices by ethnic origin group for Hispanics, Asians and all others, by immigrant generation and education group. We then formed a weighted average of these matrices for each generation and each education group, with weights equal to the ethnic shares in each generation. For first-generation immigrants we used the educational distribution of recent immigrant flows and base changes in the ethnic shares of subsequent generations on fertility differences between ethnic groups. These weighted average matrices were then used to project educational mobility in the analysis. To estimate the transition matrices for Hispanics and others, we used educational outcomes for all children age at least 21 years, and born after 1960, in every GSS since 1972. For Asians, sample sizes were quite small, so we had to consider all children age at least 21 years and born after 1950. Even so, numbers were small, so we assumed that their dropout rates at each level of education were proportional to the corresponding dropout rates of others, with a different constant of proportionality estimated for each level of parental education. From these fitted dropout rates, we estimated the distribution of educational outcomes. The resulting matrices by immigrant generation, ethnicity, and educational attainment of parents were used as described in the preceding paragraph. We also used an alternative procedure to estimate these transition matrices in which we weighted the sample in various ways. (1) Early-born cohorts are
OCR for page 356
--> represented in many different years of the survey, while later-born cohorts are represented fewer times (only once in the extreme case). We therefore weighted by the inverse of the number of times a cohort would be represented in the combined surveys. (2) Blacks were oversampled, and we have weighted to remove this effect. (3) Adults in households with more adults present are less likely to be represented, given our procedures, so we weighted by the number of adults. The various weighting procedures did not change the results in any important way. The estimated distributions of each generation's education by the education of the original immigrant are shown in Figure 7.A1. It can be seen that there is substantial upward educational mobility for the children of low- and medium-education parents, and by the fourth generation (the great-grandchildren of immigrants), the implied educational distributions are essentially identical, regardless of the educational status of the original immigrant. Although in general it can be said that each generation is better educated than the one preceding it, an important exception is seen in the educational attainment of the grandchildren of immigrants. Grandchildren (third-generation immigrants) whose immigrant forebear had at least a high school education are less well educated than their parents (second generation immigrants). Compared with the general population, we find greater upward educational mobility for the children of first-generation immigrants. However, educational mobility for children of the second-generation immigrants is much lower than that found in the general population. It would not be right to use the transition matrix estimated for U.S.-born children of immigrants to project the educational attainment of all the foreign-born children of immigrants, since they will have different English language skills and educational backgrounds. For a foreign-born child who arrives in the United States at a young age, this transition matrix is probably appropriate; but for one who arrives as a teenager, it would be much less so. Inspection of transition probabilities estimated from the GSS for foreign-born children shows that these have had somewhat less upward educational mobility than the transition matrices for U.S.-born children of immigrants would suggest. Presumably the ultimate educational attainment of foreign-born children will depend on their age at arrival in the United States. In order to take this into account, we have modified the educational transition probabilities that were estimated for first-generation children. For foreign-born children arriving at ages up to 12 years, we retain the estimated second-generation probabilities. From age 12 on, for children of low- or medium-education parents, we assume the probabilities of upward mobility decline linearly to 0 at age 20. For children of parents with higher education, we leave the probabilities as estimated. In this way we construct different transition matrices for foreign-born children, and for the U.S.-born children of foreign-born parents. Figure 7.A1 shows that we project that 52 percent of the grandchildren of immigrants will attain more than a high school education (> HS) and a higher
OCR for page 357
--> Figure 7.A1 Educational attainment of descendants of immigrant by educational attainment of original immigrant. percentage (67) of the great-grandchildren. Currently, about 52 percent of the total U.S. population aged 25 to 34 years has more than a high school education (U.S. Statistical Abstract, 1995, Table 240:158). By coincidence, this figure is identical to our projection for the grandchildren of immigrants. It is somewhat lower than our projection of 67 percent for the great-grandchildren of immi-
OCR for page 358
--> grants. These figures are not really comparable since our projections refer to educational distributions 30 to 40 years in the future rather than the current levels. Average educational attainment in the U.S. has been rising and it is reasonable to assume that significantly more than 52 percent of the great-grandchildren of nonimmigrants will attain more than a high school education. Indeed, our analysis of the intergenerational educational matrices for the nonimmigrant population indicates a long-run educational distribution in which 72 percent of the population attains more than a high school education. Appendix 7.B Net Present Values for Immigrants and Natives, by Scenario Tables 7.B1 to 7.B3 present data on the net present values for immigrants, by age of arrival and for native-born residents, in the third and later generations, at current ages, for various alternative assumptions. TABLE 7.B1 Less Than High School Education Immigrant Age at Arrival Scenario 0 20 21 40 70 Baseline 60 33 7 -141 -166 2% discount 298 228 181 -192 -181 4% discount -9 -12 -29 -104 -154 6% discount -43 -27 -36 -59 -133 8% discount -44 -24 -29 -35 -117 100% taxes 87 52 23 -160 -167 100% benefits 33 15 -8 -123 -166 Balance budget now 51 25 -0 -140 -161 Never balance budget -56 -109 -129 -164 -167 Never balance budget + welfare reform -50 -105 -123 -143 -152 Duration 10+ taxes 99 67 37 -119 -166 Duration 5-9 benefits 62 36 9 -142 -166 Welfare reform act 65 38 12 -120 -151 Elderly immigration with zero OASDHI 60 33 7 -141 -73 No emigration 2nd gen 72 50 23 -141 -166 Lower emigration 103 47 14 -174 -189 Lower bilingual education 65 35 8 -143 -168
OCR for page 359
--> Native (3rd + Generation) Current Age Difference Scenario 0 20 21 40 70 0 20 21 40 70 Baseline 92 234 -54 -134 -223 -33 -200 61 -8 57 2% discount 362 487 85 -190 -245 -64 -259 96 -2 65 4% discount 12 147 -72 -92 -204 -21 -159 43 -12 50 6% discount -32 79 -60 -42 -173 -11 -106 24 -16 39 8% discount -39 51 -46 -17 -148 -6 -75 17 -18 31 100% taxes 124 266 -47 -155 -224 -37 -213 70 -5 57 100% benefits 61 202 -61 -112 -222 -28 -187 53 -11 56 Balance budget now 83 229 -58 -129 -213 -32 -203 58 -10 53 Never balance budget -41 82 -183 -164 -225 -15 -191 55 -0 57 Never balance budget + welfare reform -41 82 -183 -164 -225 -9 -187 60 21 72 Duration 10+ taxes 92 234 -54 -134 -223 7 -167 91 15 57 Duration 5-9 benefits 92 234 -54 -134 -223 -30 -198 63 -9 57 Welfare reform act 92 234 -54 -134 -223 -27 -196 66 13 72 Elderly immigration with zero OASDHI 92 234 -54 -134 -223 -33 -200 61 -8 150 No emigration 2nd gen 92 234 -54 -134 -223 -21 -184 77 -8 57 Lower emigration 92 234 -54 -134 -223 11 -186 68 -40 34 Lower bilingual education 92 234 -54 -134 -223 -27 -199 62 -9 55
OCR for page 360
--> TABLE 7.B2 High School Education Immigrant Age at Arrival Scenario 0 20 21 40 70 Baseline 92 146 126 -32 -225 2% discount 353 391 351 -68 -248 4% discount 10 70 58 -6 -205 6% discount -35 21 16 21 -172 8% discount -40 8 5 31 -146 100% taxes 123 175 152 -48 -226 100% benefits 61 117 100 -15 -224 Balance budget now 83 138 118 -29 -216 Never balance budget -26 -2 -16 -56 -227 Never balance budget + welfare reform -21 4 -9 -47 -203 Duration 10+ taxes 133 184 161 -0 -225 Duration 5-9 Benefits 94 138 117 -44 -225 Welfare reform act 98 152 132 -23 -201 Elderly immigration with zero OASDHI 92 146 126 -32 -142 No emigration 2nd gen 104 164 142 -31 -225 Lower emigration 145 185 159 -44 -261 Lower bilingual education 97 149 128 -31 -227 TABLE 7.B3 More Than High School Education Immigrant Age at Arrival Scenario 0 20 21 40 70 Baseline 117 288 333 132 -149 2% discount 395 594 641 116 -163 4% discount 26 174 211 140 -137 6% discount -28 80 107 140 -116 8% discount -37 44 64 131 -101 100% taxes 152 332 379 119 -150 100% benefits 82 245 287 144 -148 Balance budget now 108 281 326 138 -142 Never balance budget -4 127 173 101 -150 Never balance budget + welfare reform 1 130 176 107 -119 Duration 10+ taxes 162 352 407 200 -149 Duration 5-9 Benefits 120 294 343 147 -149 Welfare reform act 122 292 335 138 -118 Elderly immigration with zero OASDHI 117 288 333 132 -54 No emigration 2nd gen 129 307 351 132 -149 Lower emigration 177 364 418 149 -170 Lower bilingual education 123 293 338 134 -150
OCR for page 361
--> Native (3rd + Generation) Current Age Difference Scenario 0 20 21 40 70 0 20 21 40 70 Baseline 171 342 182 18 -209 -79 -195 -56 -49 -16 2% discount 495 637 394 -19 -230 -142 -246 -43 -49 -18 4% discount 61 228 115 43 -190 -51 -158 -57 -49 -15 6% discount -10 130 67 68 -160 -24 -108 -51 -47 -12 8% discount -28 85 47 75 -136 -13 -77 -42 -44 -10 100% taxes 214 383 205 -1 -210 -91 -208 -54 -47 -16 100% benefits 129 300 158 37 -208 -68 -183 -58 -52 -16 Balance budget now 161 336 177 23 -197 -78 -198 -59 -52 -19 Never balance budget 31 182 44 -16 -210 -57 -184 -60 -40 -16 Never balance budget + welfare reform 31 182 44 -16 -210 -51 -178 -54 -31 7 Duration 10+ taxes 171 342 182 18 -209 -38 -157 -21 -18 -16 Duration 5-9 Benefits 171 342 182 18 -209 -78 -203 -65 -62 -16 Welfare reform act 171 342 182 18 -209 -74 -190 -50 -41 7 Elderly immigration with zero OASDHI 171 342 182 18 -209 -79 -195 -56 -49 67 No emigration 2nd gen 171 342 182 18 -209 -67 -178 -40 -49 -16 Lower emigration 171 342 182 18 -209 -26 -156 -23 -62 -52 Lower bilingual education 171 342 182 18 -209 -74 -194 -54 -50 -18 Native (3rd + Generation) Current Age Difference Scenario 0 20 21 40 70 0 20 21 40 70 Baseline 245 442 503 244 -191 -128 -154 -170 -112 42 2% discount 621 780 846 235 -211 -226 -186 -205 -118 48 4% discount 106 303 354 246 -173 -80 -129 -143 -107 37 6% discount 9 174 212 236 -145 -37 -94 -105 -95 28 8% discount -18 115 143 217 -122 -19 -70 -79 -86 21 100% taxes 298 494 557 231 -192 -146 -162 -179 -111 42 100% benefits 192 391 448 257 -190 -110 -146 -161 -113 42 Balance budget now 234 437 498 255 -176 -126 -157 -172 -117 35 Never balance budget 95 273 334 199 -193 -100 -146 -161 -98 43 Never balance budget + welfare reform 95 273 334 199 -193 -94 -143 -158 -92 73 Duration 10+ taxes 245 442 503 244 -191 -82 -90 -96 -44 42 Duration 5-9 Benefits 245 442 503 244 -191 -124 -149 -160 -97 42 Welfare reform act 245 442 503 244 -191 -122 -151 -167 -106 73 Elderly immigration with zero OASDHI 245 442 503 244 -191 -128 -154 -170 -112 137 No emigration 2nd gen 245 442 503 244 -191 -115 -136 -152 -112 42 Lower emigration 245 442 503 244 -191 -67 -78 -84 -95 20 Lower bilingual education 245 442 503 244 -191 -123 -149 -165 -110 41
OCR for page 362
--> Appendix 7.C Discussion Of Data Sets For The Study Of Fiscal Impacts Most analyses of the use by immigrants of government entitlement programs are based on one of three data sets: the Public Use Microsample (PUMS) of the decennial U.S. Census of Population, the Survey of Income and Program Participation (SIPP), or the Current Population Survey (CPS). Each has advantages and disadvantages. PUMS is the largest sample, containing data on 12.5 million individuals in 1990, of whom about 9 percent, or about 1 million, are foreign-born. Because special efforts were made to include responses from illegal immigrants, it is possible that these are reasonably well represented in the sample. The large size of the PUMS means that estimates have relatively small sampling error and that it is possible to disaggregate by characteristics of immigrants such as education, time since arrival, country of origin, and place of residence in the United States. However, PUMS contains fewer of the necessary programmatic data items compared with SIPP or the March CPS, is updated only every 10 years along with the decennial census, and has limited possibilities for intergenerational analysis. Compared with the decennial census and CPS, SIPP has a smaller sample size, and only by combining panels is it possible to get enough respondents to include an acceptable number of immigrants. In work for this report, Peter Brandon combined seven SIPP panels, 1986-93, to obtain a total sample size of 290,000 individuals, including 33,000 foreign-born. SIPP distinguishes foreign-born from native-born respondents, and for children up to age 15 reports whether their parents are immigrants. For older children, it does not provide this information. The real strengths of SIPP are that it contains more accurate monthly data on program participation and expenditures and richer information on wealth and income sources, than either PUMS or the March CPS. CPS, starting in 1994, supports separate study of first and second-generation immigrants and the remainder of the population. The sample size is about 150,000. Combining the 1994 and 1995 surveys yields about 300,000 respondents, with about 30,000 foreign-born. Like SIPP, CPS provides rich detail about program use, but these data are thought to be less reliable than those in SIPP. For our national-level longitudinal analysis, we have relied largely on CPS, viewing it as a good compromise between the sample size of PUMS and the detail of SIPP. Additional work for the panel pools seven waves of SIPP as a basis for assessing the national work done using CPS. The case study for California derives program participation rates from the March CPS but uses average costs per participant from other sources. The New Jersey case study was based on PUMS.
Representative terms from entire chapter: