CHAPTER 5 EFFECTS OF IMMIGRANTS UPON THE PUBLIC COFFERS Taxpayers are hurt by having to pay more for social services. Roger Conner, Federation for American Immigration Reform (FAIR) We spend millions every month supporting people who are not supposed to be here. Leonard Chapman, when Commissioner of the Immigration and Naturalization Service (INS) INTRODUCTION It is frequently said that immigrants to the United States burden taxpayers by heavy use of welfare services. A 1986 poll showed that "47% of Americans felt that most immigrants wind up on welfare" (CBS/New York Times, July 14, 1986, p. 1). Often this charge is laid against illegal immigrants, as in the head quotations above, in which case the charge seems especially grave because the illegals are not "entitled" to the services they receive, unlike legal immigrants. Both logic and compelling empirical evidence, shown in Chapter 15, demonstrate irrefutably that this charge against the illegals is preposterous; even if they should wish to do so, the illegals could not take much advantage of welfare services because of fear of being apprehended. Therefore, the question to be dealt with here is whether legal immigrants impose a welfare burden upon natives. This chapter first adduces excellent Census Bureau data to show that legal immigrants do not impose a burden upon natives by their use of welfare services such as medical care, unemployment pay, food programs, aid to dependent children, retirement payments, and schooling expenditures. Rather, immigrant families use less than their share of such services. This is due largely to the age composition of immigrant families, who typically are adults arriving in the U.S. in the early prime of their working lives, with relatively few children. (Chapter 6 shows, however, that even compared to natives with the same ages and educational characteristics, immigrants do not use disproportionate amounts of such services.) The chapter then adduces other good data on the earnings and tax payments of immigrants which show that the average immigrant family pays more taxes than does the average native family. The two sets of data, covering welfare use and taxes, are then combined in order to arrive at a combined estimate of the effect of immigrants upon natives through the public coffers. This combined welfare-and-tax result shows that immigrants yield a benefit to natives of sizable magnitude. Of course this effect is only part of the overall effect of immigrants, the other aspects of which are addressed in the rest of the book. But this part of the picture, showing that immigrants make a large net contribution to natives through the public coffers, is not only important but also reinforces most of the other economic impacts of the immigrants.0.1 CONCEPTUAL FRAMEWORK Individuals and societies usually have assumed that the group benefits by adding unattached persons in the prime of life. The young and the strong produce more than they consume; they support and protect the aged and the dependent children. More recently this phenomenon was discussed by Neisser (l944), and then formalized by Modigliani (l966) and Samuelson (l975), who argue that a faster rate of population growth leads to a higher rate of saving because there is a higher proportion of young workers if population growth is faster, and young workers save in order to consume while in retirement. In an ongoing growth system, the saving is in effect a system of transfers in each generation from the workers to the retired and the children; Samuelson (1975) compared this process to a chain-letter scheme, but unlike chain-letter schemes, there is no constraint that necessarily brings this process to a halt. Though the context of Modigliani's and Samuelson's discussions was the private financial system, much the same principle holds for the public tax-and-transfer system, which has obvious implications for the Social Security system. If one adds to the analysis the transfers to children as well as to the retired, the analysis is less clearcut; Arthur and McNicoll (l976) made this criticism of Samuelson. And indeed, in some societies such as Germany, the ratio of social transfer payments to children relative to the transfers to the retired is much larger than in the U.S.; this could invalidate the above argument. Ultimately, it is an empirical question whether, given a particular society's transfer system, an additional baby implies on balance a lower or higher proportion of a worker's income transferred to dependents outside his/her family through taxes than if the baby is not born. The Neisser-Modigliani-Samuelson reasoning applies with even greater force to immigrants than to new births, because the newcomers are typically young men and women who arrive without elderly dependents1 and with a relatively small number of children already born. The family rather than the household or the individual is the unit of analysis in this chapter, for reasons given in the afternote to the chapter concerning the SIE Survey and Sample used. PUBLIC SERVICES USED BY IMMIGRANTS AND NATIVES The aim of this section is to estimate the overall gross cost to U.S. citizens of admitting immigrants on average to the U.S. (Gross benefits and net effects are discussed in subsequent sections.) More specifically, we wish to estimate for the average immigrant in an entry cohort (a) the amount of each type of service used in each year after entry, and (b) the yearly total. (Some interpolation is necessary because some cohorts contain more than one year's immigrants because of the SIE survey's classification system.) The average of an entire immigrant entry cohort is the most important estimate for our purposes, because it is the relevant quantity for policy decisions about the total number of immigrants to allow in; it indicates the gross cost to natives of services to the average immigrant family. Let us look at some scraps of data from other countries before turning to the U. S. (l) Canada. In the first three years, almost 50 percent of the immigrants attended part-time or short-duration educational courses financed by the Canada Manpower Training Program, an expensive service. More important, though less direct, is the evidence on poverty among immigrant families, because it is families in poverty who especially need welfare services. With respect to the Canadian-defined "poverty line" income, a slightly higher proportion of immigrants than non-immigrants--22% compared to 21%--were below the poverty line at the end of the immigrants' first year. But by the end of the second and third years, only five percent and four percent of the immigrants were below the poverty line--to be compared with the 21% among all Canadians (pp. 55-57). Also important is the very low proportion in or near the retirement age among the immigrants--one percent or less being 65 or over, and less than five percent are 50 and over (p. 142). This suggests very low present and future burden on retirement pension funds. (2) Great Britain. Because there are proportionally so many fewer (mainly "New Commonwealth," mainly Caribbean) immigrants than natives of retirement age, immigrants receive much less in retirement pensions than natives, and pensions are the largest item in the cost of national insurance and assistance benefits (Jones and Smith, 1970, p. l04). The latter are, in turn, the biggest share of social service expenses; the figures were L15.4 and L27.6 in l96l, and L16.2 and L28.3 in l966 (p. l07). In other categories, expenditures per head for immigrants are much the same as for natives. In total, the expenditures per head for immigrants and total population were L47.5 and L6l.0 in l96l, and L52.0 and L62.0 in l966 (p. l07). Thus, immigrants are not relatively heavy users of welfare services; rather, they are relatively light users of welfare services. (3) Israel. Israel is quite atypical with respect to the supply of social services to immigrants, for several reasons. First, the immigrants from the Soviet Union and elsewhere nowadays move largely for non-economic reasons, in contract to most migrants elsewhere in the world. Second, the social services supplied--housing subsidies, special rights for the importation of autos and other goods, and acclimatization services--are quite large (though not as large as most Israelis believe) and they are hard to estimate because they vary considerably across immigrants. Third, foreign charity funds more than cover the costs of these social services, so they do not affect resident Israelis. (4) Australia. Using data on the relationship of demographic structure to government expenditure on education, social security, and welfare and health care, Tulpule (1984) calculated that migrants lower the necessary taxes paid by native Australians to finance such government expenditures. United States. From the l976 Census Bureau Survey of Income and Education (SIE, described in the afternote to the chapter), I constructed a picture of lifetime economic behavior by assuming that the information on immigrants who had been here, say, two years as of l975 describes the representative immigrant family after two years, those in the U. S. 10 years in l975 stand for the tenth year of a representative family, and so on. The services that most often catch the public eye are welfare and supplemental security payments, unemployment compensation, medical care, aid to dependent children, and food stamps. Table 5-l shows the amounts of payments received by persons who have been in the U. S. varying lengths of time. Columns l-8 show the various transfer payments to natives and to immigrant entry cohorts, with the spotlight on the cohorts ranging backward from l974 to l950. The meaning of the results for families that arrived in l976 and l975 is exceedingly unclear, because the data supposedly refer to the calendar year l975.2 Data for the cohorts that arrived in l949 or before are shown below for completeness, although they are not relevant to policy decisions concerning immigration because these older immigrants must now be seen as part of an equilibrium system to be described below, and because the effects happen so long after admission that their discounted present value is very low at admission. Their children, and the economic impacts of those children, are an important part of the total effects of this cohort, but data on their children are not available; for the same reason, the l950-l959 cohort should not be considered an important part of the sample.3 - - - - - - Table 5-l - - - - - - The mechanism of the above-mentioned equilibrium system is not immediately obvious, and requires amplification. We are concerned with the resource flows due to immigrants that affect the economic position of natives taken as a group. That is, we want to compare the consumption of natives (including consumption of services) in two hypothetical situations, one with and one without a group of immigrants. We are not concerned with legal obligations, entitlements, equity, or the consumption of immigrants. A simplified model may help. Consider a community of subsistence farmers in which there is a surplus of land and each farm produces the same output. Each family consists of three generations: a married couple, two children, and two retired adults. Children do not work until age 20, at which time they marry, have two children, and work for 20 years. At age forty the couple retire and live until age sixty. On each farm, then, there are always one working couple, two dependent children, and two dependent elderly persons, a stationary demographic system. The farm is assumed to produce no surplus or saving; all production is consumed. Consider, now, a newly married couple who move into the community. For 20 years, the production of this immigrant family not only supports the couple and their children, but also leaves a surplus that the community can tax. During this period the family is an economic benefit to natives as a result of the taxes they pay. After 20 years, when the immigrant couple retire, the family has come to have the same characteristics as an "equilibrium" native family, and the retirement consumption of the immigrant couple is paid for by their children. This illustrates how there is a one-time benefit to natives during the first years that an immigrant couple is in the community, while there is no reverse flow after the couple retire. And the nature of the obligation between the immigrant couple and their children (which leads their children to support them) has no bearing upon the consumption of natives. The situation is analogous in a modern society except that the transfers do not take place within the family system but rather through the public coffers. Column 9 in Figure 5-1 shows the totals for the five most important categories of welfare payment transfers aside from Social Security. The average family in each of the cohorts of immigrants since l950-l959 uses about the same or slightly more such services than do native families--ranging from $l37 less to $l48 more. But when Social Security4 is included, we see in column l0 that immigrants receive much less such welfare payments in total than do natives--ranging from $294 less to $823 less. Given the consistency in composition from cohort to cohort, there is no question of the statistical significance of the smaller use of welfare transfers by immigrants than natives. Expenditures on education must be considered, too. The SIE sample does not show the total number of children in the family, but it does show the number living at home. This latter number should be a satisfactory approximation for young families and for children aged l7 or less. Column ll shows the average number of children per family aged 5 to l7. These numbers were multiplied by the l975-76 average annual expenditure of $l,302 per student in public elementary and secondary schools5; the results are also shown in column ll. Immigrants use substantially more school services than do natives after a few years--$476 more per family for both the l960-64 and l950-59 cohorts.6 Next, we consider Medicare and Medicaid.7 Columns 7 and 8 show the proportions of the various groups that use these services. The average cost per user of Medicare was $592 in l975, and for Medicaid it was $l26. (The average cost per person over 65 was $205 for Medicaid, and a considerably larger proportion of natives than of recent immigrant cohorts are over 65, but I shall not bother to make an adjustment for this.) Special assistance to refugees--which in l980 amounted to $l,000 per person in cash plus some overhead to the assisting refugee organization, as well as some special programs--has also been excluded from the analysis, because there is no way of knowing which families in our sample were admitted as refugees. If one wishes to make a benefit-and-cost analysis specifically for refugees, it would therefore be appropriate to deduct this sum, multiplied by the average number of family members, from the present value calculated for the family, as discussed below. Refugees, however, are not considered as immigrants according to quotas and immigratic law. We are now in a position to estimate the average total cost of services used by the various cohorts, with results shown in column l2. There we see that the very recent cohorts use much less of services than do natives--for example, $97l less per family for the l973-l974 cohorts, considered together, than for natives. The amounts of services used by older cohorts are higher; they move upward and reach equality with natives some l0- l5 years after arrival. Figure 5-l plots the data on services used. A free-hand curve allows us to calculate the change in services with an additional year in the United States. ---------- Figure 5-1 ---------- Native families are used as a benchmark because if immigrants were to use services and pay taxes in amounts equal to natives, the immigrants would have no effect on natives through the public coffers; on average, taxes paid equal the cost of services provided. In the over-all cost-benefit framework employed later, personal services used by immigrants cannot simply be compared to taxes paid by immigrants because the "overhead" of public goods must also be allowed for; the comparison to natives takes care of this. Concerning illegal aliens now, the evidence is clear and strong that they do not use welfare services, both because they are young, mature, and strong, and also because they cannot use such services without fear of the law. Chapter 15 presents data on this. Conclusion The data from the "normal" cases -- Britain, Canada, and the United States -- suggest that immigrants use a below-average quantity of services, largely because of the "favorable" age and sex composition of the immigrant groups. TAXES PAID BY IMMIGRANTS Please notice the difference between the data on unconditional family earnings discussed in this section, and the studies of immigrants' earnings with as many factors as possible held constant in the large literature on that subject. Nothing is held constant in this cost-benefit analysis other than year of entry, because the aim of this chapter is to assess the unconditional impact of a cohort of immigrant families upon natives' standard of living. In contrast, the aim of the other studies is to understand the nature of the differences between individual immigrants and natives when such factors as education and age are allowed for. The broad patterns of the unconditional family earnings results, however, clearly are compatible with the results of the controlled individual comparisons. Again, we shall first consider the fairly scanty but suggestive data for Canada, Great Britain, and Israel, before going on to a solid set of extensive data for the United States. l. Canada. No directly relevant tax data are available. We do know, however, that immigrants' average income was 12-14% below that of the native-born control group after three years, and hence the tax collections would be somewhat less for immigrants than for native-borns. It is reasonable to assume, however, that immigrants' incomes relative to non-immigrants rise after that, and wipe out differences in tax payments. Also some non-negligible segment of taxes goes for defense and other centralized activities in which there are major economies of scale. Hence, even immigrant tax payments somewhat lower than natives' might more than pay for an average person's use of services, let alone the lower use of services by immigrants, who include fewer retired persons and fewer in poverty. 2. Britain. The survey data available to Jones and Smith (1970) did not allow them to estimate immigrant income-tax payments. Their best guesses were as follows: [T]here is some evidence that average income per employed New Commonwealth worker is lower than the national level, but we also know that the labour force participation rates are higher for the New Commonwealth than for the indigenous population, so total household income from earnings alone for New Commonwealth households may well not deviate much from the national average. If this were the only factor it would imply similar national insurance payments and income tax payments per head. But we also know that the dependency factor is higher for the New Commonwealth households and this would lead to greater income tax allowances and in consequence smaller payments. It is also known that there has been some tax evasion by immigrants and this has been estimated at 5 to 7 million [pounds] in l966/7 by the Board of Inland Revenue; but now that this has been identified, an attempt will presumably be made to stop it. Whilst it is not practicable to estimate the relative contributions of immigrants in the form of national insurance contributions and income tax payments, it seems unlikely that, if New Commonwealth immigrants do pay less, the difference would more than outweigh their smaller share of the social services (Jones and Smith, l970, p. 95). 3. Israel. The tax situation of immigrants to Israel is quite atypical among countries because of (a) the non-economic motives for immigration, (b) the special subsidies and tax breaks for immigrants, and (c) the large defense burden. Of the taxes that all Israelis pay (perhaps 60-65% of GNP) a very large proportion (perhaps 40%) goes to defense. These contributions to the defense burden by immigrants--in taxes, and in time spent in the army--substantially lighten the portion of the communal burden borne by native Israelis. The United States The Census Bureau's 1976 SIE survey (mentioned earlier) allows us to estimate the taxes paid by immigrants. Though we lack direct information on taxes paid, we can make a fair estimate from data on family earnings. Column 4 in Table 5-2, and Figure 5-2 based on it, show that the average immigrant family comes to earn about as much as the average native family somewhere between three and five years after entry, and after that earns more. (This finding uses an average for the l970-l973 cohorts.) This rapid approach to equality of earnings is heavily influenced, of course, by age and education composition, and especially by the absence of retired family heads among the immigrants. ------------------------- Table 5-2 and Figure 5-2 ------------------------- Inferences from earnings data about taxes paid by various cohorts of immigrants, and by the representative native family, are tenuous but necessary. A key difficulty is that different proportions of overall taxes (not just income taxes) are paid by persons of different income levels (Browning and Johnson, l979), which means that a proper calculation must be based upon the entire income distribution of a cohort, rather than just the mean.8 A bit of light is thrown on the matter by the spread of the income distributions in our data, which show somewhat smaller standard deviations relative to mean income for immigrants than for natives, as seen in Table 5-2 (in parentheses). This probably means that, for the same mean income, a group of immigrants pays somewhat less taxes than does a group of natives, due to fewer very high incomes. Nevertheless, we are not likely to go far wrong by assuming that taxes paid are proportional to mean income. The total of taxes paid by immigrants--found in columns 10 and ll of Table 5-2--are calculated as follows. The total receipts of state, local and federal governments were $77l.9 billion in l979, which we may approximate as 32.6% of gross national product ($2,368.5 billion; Economic Report of the President, l980, pp. 288, 203). The mean-taxes-paid taxpayer pays 29.1% of income; the taxes paid by the median-income taxpayer are 85% of that; that is, the rate at which taxes are paid by the person between the fifth and sixth deciles of income recipients--24.8% of income paid in taxes--is 85% of the mean 29.l% for all taxable incomes (Browning and Johnson, l979, p. 5l). So even if immigrants are more concentrated near the mean than are natives, the effect on total taxes paid will not be large. In brief, although the estimates of taxes paid by immigrants and natives are rough, the inaccuracy is unlikely to be important in this context because the incomes--and hence the amounts of taxes paid by the various cohorts and by natives--do not differ substantially. Conclusion Immigrants tend to pay quantities of taxes that exceed the costs of the public services they use. NET EFFECT ON THE PUBLIC COFFERS If we know the amount of taxes paid by immigrants and the amount of welfare services used by them, we may then compute the net balance, positive or negative, for immigrants as a group. Additionally, we can then compute their impact upon the public coffers. Table 5-3 consolidates the relevant data on services used and taxes paid. Columns 1 and 2 show the total transfer payments and services used by immigrants in various years after entry (actually, by various entry cohorts during l975) and by natives. Column 3 shows the differences between columns 1 and 2. If we make the reasonable assumption that the average family just pays for the average family's services used--an assumption that says no more than that government receipts equal the sum of government expenditures for various purposes--then column 3 indicates the net balance of immigrants with respect to services alone. In other words, it indicates the amount by which services used by immigrants exceed or fall short of the amount an average taxpayer9 pays for that. ---------- Table 5-3 ---------- Turning from services used to taxes paid, we first deal with government expenditures that are invariant to the number of people, or that at least are pure public goods in the sense that the services they render to any person are independent of the services others obtain from them. The government expenditures that seem clearly to fit into this category are national defense ($85,552 million in l975), international affairs ($6,922 million), and general science, space, and technology ($3,989 million), out of total outlays of $326,l85 million, or ($96,463/$326,l85 =) 29.6% (p. 284, Economic Report of the President, l980). Federal government expenditures were ($356.8 /$532.8 =) 67% of total government expenditures (p. 288), so the public goods categories constitute (67% x 29.6% =) 20% of all taxes. Hence 20% of taxes paid by immigrants are shown in column 6. Next we consider other taxes. The differences between the amounts of taxes paid by immigrants and by natives are shown in column 8 of Table 5-3. To the extent that immigrants pay less or more than natives, immigrants are taking from, or giving to, natives through the public coffers, after allowing for the services they use. We may summarize the logic of the tax calculation as follows: Benefit to natives = (taxpaid services used by the average native family less taxpaid services used by the average immigrant family) + (20% of taxes paid by the immigrant family as a contribution to expenditures on public goods) + 80% of the difference between taxes paid by the average native family and the average immigrant family, a negative quantity in some cases). A calculation is also shown in column ll of Table 5-3 for the less realistic but more "conservative" assumption that there is no public-good element in immigrants' taxes that immediately lowers the tax burden of natives (or increases benefits from the tax-paid services). We are now in position to sum the effects on a year-by-year basis. This summary is shown in column l0 of Table 5-3. There we see that in every year following entry (until the immigrants themselves retire, at which time their own children are supporting them through the Social Security and Medicare systems) immigrants benefit natives through the public coffers. And a calculation of the net present value of the stream of differences shows that immigrants are a remarkably good investment at any conceivable rate of discount, if we consider these partial effects only. At a 3% discount rate--roughly the long-run riskless rate adjusted for inflation in the Western World--each immigrant family was worth about $20,600 to natives around l975, to be compared with the mean yearly native family earnings of $ll,037; at 6% the present value would be about $l5,800, while at 9% it would be about $l2,400, in 1975 dollars. When thinking about the totality of the economic effect of immigrants upon natives, one must take into account, in addition to the calculations discussed here, both capital dilution (negative effect on natives) and productivity (positive effect on natives). Estimation of these latter effects as well as the totality requires a model of the economy at large, plus some key parameter estimates such as the proportion of the returns to existing capital that are captured by immigrants. But based on the integrated model discussed in Chapter 10, one may conclude that the average immigrant is an excellent investment for taxpayers on almost any reasonable set of parameter estimates. For comparison, let us examine a set of data that comes from a smaller sample drawn in a less rigorous fashion, l98l interviews of Jewish immigrants from the Soviet Union (R. Simon and J. Simon, 1985). The separate amounts received in public aid, and paid in taxes, are not immediately comparable to the SIE data, and we did not draw a separate comparison sample of natives because this was only one part of a more general study of these immigrants. But the flows to and from the public coffers may be compared to see the overall effect on natives through the public coffers. The flows to the immigrant families in public-aid transfer payments, educational services, and the "block grant" of assistance to refugees via the refugee organizations (the latter also spent for Vietnamese and other refugees), are shown in lines 1-3 of Table 5-4, with the total shown in line 4 for the first six months and then for twelve-month periods after that. The average immigrant family's income is shown in line 5, the taxes calculated from that income are shown in line 6, and the net flows from and to the public coffers are shown in line 7. We see that in the first six months there is a substantial net flow from natives to immigrants, but the flow quickly reverses and by the end of the second full year the flows from immigrants to natives have become substantial. If one calculates the present value of these flows as one would other investments, at a 3% "real, riskless" rate as well as at 6% and 9%, one gets sums of $20,593, $l7,692, and $15,262 (in l98l dollars). --------- Table 5-4 --------- DISCUSSION l) A major limitation of these data is that inferences about life-cycle behavior are drawn from a cross-section; the difficulty is widely understood. Furthermore, the sub-groups observed in the cross-section--cohorts that came to the U.S. in different years--differ from each other in relevant characteristics. Yet the analysis assumes that all the cohorts are proxies for some portion of a single cohort's economic life-- a cohort which has not yet arrived and whose composition is not yet known. The cohort-wide data shown here should provide fairly accurate general guidance, however, partly because average education has not changed greatly from cohort to cohort (though age and ethnic composition have changed). Also, the general logical consistency in the pattern of the various cohorts substantiates that the cohorts that have arrived in the past are reasonably similar, and offer a reasonable basis for prediction of the effects of future cohorts. 2) In principle it is possible to isolate the effects of various characteristics such as education, age, and perhaps ethnicity in past cohorts, and then to estimate the economic effects of a cohort expected to have a particular mix of these characteristics. But there is no basis on which to make such disaggregated estimates for a synthetic cohort until we know the composition of the future cohort -- it would not be worthwhile to do so, and unless we would know that it would be very different from past cohorts, of course. (Without further investigation, however, we can say that any policies leading to higher education and lower age of immigrants will be beneficial to natives.) 3) The contemporary public-coffers benefits to natives from immigrants derive overwhelmingly from the age composition of natives rather than from behavioral characteristics of immigrants; this may be seen in the importance of the Social Security effect. 4) Some opponents of immigration (and even some friends of immigration) have said, roughly, "Yes, the United States received much larger numbers of immigrants, relative to the number of natives, in the past than it does now, and absorbed them successfully. But the conditions were different then. There was no program of social welfare and public support that attracted immigrants and that was costly for natives to pay for." In fact, the argument actually points in exactly the opposite direction, because the contemporary structure of social services and taxes-- which involves larger amounts of support to the needy, and therefore larger burdens upon the taxpayers, than in the past--is on balance a net beneficiary from immigrants rather than a net contributor to them. That is, more immigrants mean a lighter rather than a heavier burden upon natives because immigrants are, as we have seen, net contributors to the system due to their age and family composition. And this contribution is greater than it was a century ago, say, because fewer services were then being provided that the immigrants helped to pay for. So--perhaps ironically--the very phenomenon that causes many Americans to object to immigration turns out, upon investigation, to be a strong argument for increased immigration. 5) About illegals: Chapter 15 shows that illegals use practically no welfare services, while about three quarters pay Social Security and income taxes. This means that illegals contribute heavily to natives in this regard; the native population is "exploiting" the illegals in this fashion, rather than the opposite. 6) Immigrants chosen for their economic characteristics rather than for their being relatives of families already in the U. S. surely use far less welfare services than do relatives. Evidence for this comes from a survey of General Eligibility Migrants in Australia: '[T]his category of migrants is not a heavy user of, nor does it generally have unusual need for government services" (Northage & Associates, 1983, p. 63). 7) Different performance of different ethnic groups with respect to the public coffers is frequently asked about. Chapter 6 shows that family structure overwhelmingly dominates all other variables in explaining welfare receipts aside from old-age programs; unmarried women with several children receive much more than do other types of families. And no group of immigrants is very different from others in this respect. In human-capital adjusted regressions, Mexicans receive considerably less in services--$436 per year--than do other immigrant families on average. On the earnings-and-taxes side, Mexicans earn considerably less than do other ethnic groups. But the low amount of services used offsets this shortfall. Aside from this, there is not much noteworthy about the data for the various ethnic groups. 8) Refugees may be expected to take longer to establish themselves, and therefore to have higher rates of welfare usage for longer periods of time, than other immigrants. Therefore, it is of interest that as of mid-1983, the proportions of refugees (largely Southeast Asians) arriving from 1980 to 1983 and receiving cash assistance as of the survey date were: arrived fiscal year 1980: 7%; FY 1981: 13%; FY 1982: 26%; FY 1983: 24% (Church World Services, 1983, p. 38). The proportions with full- time employment, as reported by the sponsors and by the refugees, respectively, were: Fiscal year 1980: sponsors, 74%, refugees, 65%; FY 1981: 65%, 55%; FY 82: 55%, 39%; FY 83: 44%, 30% (Church World Services, 1983, p. 25). 9) The ongoing debate over Social Security is agonizing. Every proposed policy inflicts pain because it boils down either to a decrease in benefits to the elderly or an increase in taxes upon working persons. And so it must be, as long as the numbers of elderly and of workers remain unchanged. It is unthinkable that the number of elderly be reduced by our health policies. There remains, then, only one condition that can lessen the tax burden upon working persons without reducing benefits -- a larger number of young working persons. Having more children will help, but not until a couple of decades hence. Only immigration provides immediate help. As soon as the average immigrant begins working -- which generally is quite soon after arriving in this country -- the immigrant begins to contribute to the Social Security coffers. The immigrant's own eventual receipt of Social Security benefits, decades down the pike, does not offset these immediate benefits to natives, as we have seen. The benefit to natives stems from the difference in age composition between the population of present residents of the U. S., and that of each cohort of immigrants. Immigrants tend to move when they are young and near the start of their work lives, as Chapter 3 shows. For example, among immigrants who arrived in l967-l973, 4.3% were aged 60 or over, whereas 14.2% of the U. S. population at that time were 60 or over, and the proportion is even higher now. And whereas 25.7% of the U. S. population were in the early prime labor force ages 20-39, 46.4% of immigrants were in that age bracket. Moreover, even that small fraction of the immigrants who are elderly are not eligible to receive Social Security. Therefore each cohort of new immigrants contributes substantially to reduce the Social Security burden of natives, in proportion to its numbers. The anti-immigration organizations attempt to talk away this benefit of immigrants by saying that the total dependency burden of the U. S. population is not increasing because as the number of aged persons increases the number of dependent children decreases. For example, "Over the next generation, each member of the labor force will support fewer non-workers, even with less immigration than at present. There will be more older non- workers to support, but fewer young ones" (THE DATA, The Environmental Fund, July, l982, p. l). But dependent children have a very different impact upon the taxpayer than do the elderly. Parents pay most of the cost of raising their own children in this country, and do not receive the large child allowance payments some European countries give. Additionally, the living costs of a child for a year are far less than the living costs of an aged adult; this is obvious when one reflects on the comparative costs of housing, medical care, and transportation; children's public school education expenditures do not much alter the overall picture. So to lump together dependent children and dependent adults on a one-to-one basis obscures the matter. As to whether U. S. natives must pay the piper for this benefit from immigrants when the young immigrants get older and themselves receive Social Security. The answer is "no," for two reasons discussed earlier: First, the impact of this year's immigrants on Social Security say thirty years from now properly has little weight in the overall assessment, because a dollar to be received or paid out 30 years from now is worth much less now when discounted at even a modest rate. Second and more important is the mechanism analyzed above with a simple three-generation model: By the time the immigrant workers retire and collect Social Security, they typically have raised children who are then contributing taxes to Social Security and thereby balancing out the parents' receipts, just as is the case with native families. In this way there is a one-time benefit to natives because the immigrants arrive without a generation of elderly parents who might receive Social Security. Immigrants are not a complete cure for the Social Security problem, simply because the numbers of immigrants that might conceivably be allowed by any likely U. S. policy are limited. But the extent to which immigrants can be at least a partial remedy is easily underestimated. One reason that Social Security taxes per worker are as high as they now are is that -- contrary to the impression that the anti-immigration lobby tries to leave -- the country has admitted very few immigrants in recent decades, far less as a proportion of the population than in the decades around the turn of the century (See Chapter 3). If additional immigrants to the extent of a quarter of our labor force had been admitted, the Social Security tax per worker would be almost a quarter lower than it now is. The data showing immigrant fertility not to be high relative to native fertility may please some who worry (without cause) about a "population explosion". But in this context it should be noted that the fewer children immigrants have, the fewer persons there are in future decades to support their parents and others. This issue would be less pressing if the United States were to move toward a self-financed retirement system and away from a pay-as-you-go Social Security System, but by the same token immigrants would thereby lose some of their attractiveness to natives because there would be less current welfare burden for them to help pay for. Even if one favors replacing the Social Security program with private pension plans, immigration would still be of benefit, because the obligations to all those persons who are already covered by Social Security will surely be honored by the U.S. government, and immigrants would be contributing to the liquidation of that body of obligation. SUMMARY The effects in the years after arrival of an average immigrant family upon the income of an average native family, through the cost of welfare payments and schooling as well as through the benefits of the taxes paid by immigrants, are estimated from the 1976 Survey of Income and Education. From the time of entry until about l2 years later, immigrants use substantially less than do native families of such public services as welfare and unemployment compensation payments, food stamps, Medicare, Medicaid, and schooling for children, largely due to less use of Social Security because of the youthful age of immigrants. Later, immigrant use becomes roughly equal to that of natives. By the time the immigrant family retires and collects Social Security, it typically has raised children who are then contributing taxes to Social Security and thereby balancing out the parents' receipts, just as happens with native families. In this way there is a one-time benefit to natives because the immigrants arrive without a generation of elderly parents who might receive Social Security. After about three to five years, average immigrant families earn as much as average native families and thereby pay as much in taxes as do native families; subsequently they earn more and pay substantially greater taxes. The net balance of the two forces--taxes paid and services received--has a positive effect on natives in every year. That is, immigrants contribute more to the public coffers than they take from them. When looked at by natives as an investment, similar to such social capital as dams and roads, an immigrant family is an excellent investment worth somewhere between $l5,000 and $20,000 in 1975 dollars, even calculated with relatively high rates for the social cost of capital. (This is to be compared with mean yearly native family earnings of about $ll,000 in that year.) Data from Great Britain, Canada, and Israel confirm the key data concerning the United States: Immigrants use little services, and pay substantial taxes, relative to natives. M/D/266, #2 86-82 Coffers4 12/10/87 AFTERNOTE TO CHAPTER 5 ABOUT THE SIE SAMPLE The sample data used for the estimates made in this chapter also are the basis for estimates and analyses in other chapters. Therefore the data will be discussed in some detail. The l976 Survey of Income and Education (SIE), carried out by the U. S. Bureau of the Census, gathered information on the l975 incomes and use of social services by 158,500 households, stratified in such a manner as to include more-than-proportional numbers of households with children living in poverty, and is available on a set of tapes. The survey coverage is good, because only 7,300 households refused interview. All households with foreign-born heads (about 15,000) constitute the main sample used in this work. A random subsample of about 11,000 native-American households drawn from the entire sample is used for comparison purposes. The native sample was drawn with four systematic starts, so that the means of the sub-samples may be compared as a quick check of sample variability. Persons in private and public institutions such as nursing homes and hospitals were not included in the SIE survey. Puerto Ricans are not included in the immigrant group because they are U.S. citizens. As a separate group they may be of interest, but they are not relevant to the social decisions to which this book pertains. It is not known how many illegals are in the sample of immigrants, because no question indicates such status. But given the relatively small numbers compared to the total of immigrants in the U.S., and given the relatively small number of persons of Mexican and Caribbean origin in the sample, it is doubtful that illegals could in any case have much of an effect in the overall results. Similarly, there is no question in the questionnaire that indicates whether a person is an "economic" immigrant, a "family completion" immigrant, or a refugee. But one may presume that many of the Cubans, and many of the Russians and Poles who entered subsequent to l965 or l970, are refugees. Families with female immigrants married to male natives are excluded from the main analyses for the following reasons: Imagine a male immigrant who marries a female immigrant and has children. It is reasonable to view the effects of the couple plus their children as the full effects of the male's and female's immigration. Now imagine that the same male and female had instead married natives who would otherwise have married each other, and that both families are now counted in the immigrant sample. Though there is no more total effect on natives' income than before, on average, the results would show twice as much effect, which would be double-counting. We can get rid of most of the misleading double-counting by counting only male- immigrant-head and single-female-head families.10 (Because the numbers of male and female immigrants are not the same--the number of female immigrants in the childbearing ages, all in all, is somewhat greater than the number of males--the procedure is not perfect. But the numbers are close enough for most practical purposes.) The family rather than the household or the individual seems to be the appropriate unit of analysis, for several reasons. It is on the basis of family needs that public welfare, aid to families with dependent children, and similar transfers are received. Also, children are attached to both parents where present, rather than to just one of them. Furthermore, the earnings of the various members tend to be pooled by the family in consumption as well as for assessment of income taxes. The household (though in most cases it is identical to the family) is not such a long-lasting unit with respect to other nonrelated earners, nor is income so likely to be pooled within it, or transfers to be based upon it. Associated with this reasoning is the decision to consider single immigrant women as a unit of analysis. A woman who immigrates single and then marries an immigrant disappears as an independent economic entity, and her impact is then through the family unit. For a single woman, the opposite is the case, and hence for unmarried adults, both males and females are included as separate families, without there being double-counting. The status of persons born abroad who spend some or much of their pre-adult years in the U. S. poses interesting questions. The most important issue concerns whether these persons and their families should be included in overall calculations for the immigrant family population. On the one hand, the presence of these persons is the direct consequence of social policies to admit immigrants; on the other hand, these persons are not likely to show the effects of the immigration experience. Luckily for the clarity of our analysis, however, practically none of these persons appear in the cohort samples for the first ten years or so after entry, because there is not sufficient time for them to become heads of families, and these newer cohorts dominate present-value calculations of the impact of immigrants. 86-82 Cofffers4 1/15/87 FOOTNOTES 0.1In a paper received too late to be thoroughly considered here, Jensen (1987) reported a careful study of 1960, 1970, and 1980 census data showing an increase in poverty among new immigrants, while poverty among natives fell. Poverty relates to both services used and taxes paid. It is not clear, however, what light that trend throws upon the state of transfers between immigrants and the public coffers, or what the explanation is. 1If the aged do come, they do not qualify for public retirement funds. 2George Gray of the Census Bureau, who was in charge of the SIE Survey, said that there is no obvious explanation for the reporting of l975 earnings and services used by the cohort who entered in l976. The earnings could be earnings abroad, but the services could not be so explained. The likeliest explanation in his view is confusion on the part of the interviewers; the interviewers were not instructed specifically on this matter. 3Additionally, discounting for a quarter century and more reduces the meaning of contemporary events attached to these persons. 4Again, the fact that Social Security is "earned" is not relevant in this context. Whether a payment is "earned" or is "charity" refers to the legal and ethical obligations between persons. Our focus here is only on whether certain flows do or do not take place. 5U.S. Department of Health, Education, and Welfare, Statistics of Public Elementary and Secondary Day Schools, Fall l975, Table 2. 6Persons in the institutional population among natives and within the various immigrant cohorts are not included in the SIE sample. We can be sure, however, that such persons are found in larger proportions among natives because the elderly constitute a large part of that subpopulation. But given that less than one percent of the population is in institutions, implying an average annual expenditure of perhaps $60 per person in total (even less for public expenditures), the matter may safely be neglected. 7U.S. Department of Health, Education, and Welfare, Publication No. (SRS) 76-03l50, NCSS Report B-l (l2/75), Medical Assistance (Medicaid) Finances Under Title XIX of the Social Security Act, December l975. 8The distribution can be partially estimated from these data (and it is done in Chapter 13) but unfortunately the raw data on the original data are coded to "$50,000 or more" for individuals and "$75,000 or more" for families despite that the raw data included actual figures for the highest-income families. (Top- coding here is intended to avoid disclosure; top-coding was also done for data on assets such as value of home, and liquid assets). Therefore, it is conceivable that two groups with the same calculated mean income from our data would actually pay different amounts of taxes (a) because distributions of families with the same mean income can pay different amounts of taxes; and (b) due to top-coding, the actual means can be different though the calculated means are the same. 9The average immigrant may here be considered to be like the average taxpayer, subject to adjustments to come. 10Comparisons have been made of married-couple families with (a) only the male an immigrant, and (b) only the female an immigrant, to see whether including only the males introduces bias. Direct comparisons holding entry-date and/or age of immigrants are inappropriate because the female immigrant's husband is, on average, several years older than the male immigrant. With this in mind, the comparisons do not indicate any major difference. Perhaps more convincing, adding the families with the female immigrant only to the general sample does not much affect the results. 86-82 Coffers4 11/4/86 REFERENCES W. Brian Arthur and Geoffrey McNicoll, "Samuelson, Population, and the Transfer Effect," Mimeo, l976. Francine D. Blau, "Immigration and Labor Earnings in Early Twentieth Century America," in Julian L. Simon and Julie daVanzo (eds.), Research in Population Economics, Vol. 2 (Greenwich: JAI Press, l980). Edgar K. Browning and William R. Johnson, The Distribution of the Tax Burden (Washington: American Enterprise Institute, l979). Barry R. Chiswick, "The Effects of Americanization on the Earnings of Foreign-Born Men," Journal of Political Economy, l978, Vol. 86, No. 5, pp. 897-92l. K. Jones and A. D. Smith, The Economic Impact on Commonwealth Immigration (London: Cambridge University Press, l970). Francis Modigliani, "The Life Cycle Hypothesis of Saving, The Demand for Wealth, and the Supply of Capital," Social Research, Vol. 33, Summer, l966, pp. l60-2l7. Hans P. Neisser, "The Economics of a Stationary Population," Social Research, Nov., l944, 470-490. David North, Seven Years Later: The Experiences of the l970 Cohort of Immigrants in the United States, R & D Monograph 7l, U.S. Department of Labor, ETA (Washington: ETA, l979). David S. North and Marian F. Houston, The Characteristics and Role of Illegal Aliens in the U.S. Labor Market: An Exploratory Study (Washington: Linton and Company, March, l976). Paul A. Samuelson, "Optimum Social Security in a Life-Cycle Growth Market," International Economic Review, Vol. l6, Oct., l975, 530. Julian L. Simon, "The Economic Effect of Russian Immigrants Upon the Veteran Israeli Population: A Cost-Benefit Analysis," The Economic Quarterly, 23, August, l976, pp. 244-253 (in Hebrew). , "The Really Important Effects of Immigrants Upon Nature's Incomes," in Barry Chiswick (ed.), Conference on Immigration (Washington: American Enterprise Institute, forthcoming). Ashok Tulpele, "Effects of Alternative Migration Strategies on Government Expenditure", Committee for Economic Development of Australia, Information Paper No. P12, August, 1984 86-82 Coffers4