CHAPTER 6 HOW MUCH WELFARE AND PUBLIC SERVICES DO IMMIGRANTS (AND NATIVES) USE?0.1 The previous chapter disproves that immigrants are a burden upon natives because of the welfare services they use. The age distribution of immigrants implies less use of Social Security, by far the most expensive program among the various public services. This chapter studies whether, aside from Social Security, immigrants have a tendency to receive more welfare payments than do natives with the same general characteristics. To put the matter more technically, the previous chapter discusses the unconditional amounts of payments and services that immigrants receive, relative to the amounts natives receive, as part of the cost-benefit analysis of taxes and transfers. This chapter analyzes the effect of various characteristics-- especially the fact of being an immigrant, all else equal--upon the receipt of public welfare funds. That is, unlike the previous chapter, which was purely a cost-benefit assessment, this chapter seeks to understand the mechanism. The chapter also attempts to cast some light on how various background characteristics might be used as criteria for immigration policies so as to benefit the native residents of the country of immigration. It is crucial to keep in mind that this chapter deals only with welfare payments other than retirement allowances such as Social Security and Medicare. In an analysis that includes Social Security, immigrant families are immediately seen to receive far less on average than do native families simply because immigrants tend to come when they are young, and the few aged immigrants are not entitled to Social Security. Retirement transfers are excluded from this chapter because they differ fundamentally from welfare payments; the determinants of eligibility and use for retirement programs are mainly age rather than indigency. For similar reasons, services such as schooling are also excluded from the analysis in this chapter. Before beginning the main analysis for the U.S., let us take notice of some recent Canadian data. Table 6-1 shows that, except for Asians who arrived between 1975 and 1979, the proportions of Asian and U.K. immigrant families and singles who are below the poverty line -- and presumably receiving some welfare assistance -- are much lower than for the Canadian-born. The results for the recent Asian immigrants may partly reflect the experience of Vietnamese refugees. But in any case, the data for Asian-born families who arrived earlier suggest that in time these more recent immigrants will also have relatively low proportions in poverty. That the Asian-born families do eventually suffer less poverty than Canadian-born families is shown by the results for the same families observed in 1971 and then in 1981. Of course the low proportions of immigrants needing welfare assistance derive directly from their age distribution as well as their educational distribution. But age composition is a fundamental fact about immigrants, and must be taken into account in all thinking about immigration policy. --------- Table 6-1 --------- For general interest, Figure 6-1 shows the proportions of Southeast Asian refugees receiving cash assistance at various periods after entry into the United States. ---------------------------- Figure 6-1 -- Kaplan Chart 3 ---------------------------- CONCEPTUAL FRAMEWORK OF WELFARE INCOME RECEIPT Welfare transfer payments and other services may usefully be thought of as a kind of income, because they are "in-come" in nature, and also because there is likely to be substitution among earned income, unearned income, and transfer income. Therefore, the appropriate conceptual framework for thinking about transfer income is likely to be similar to that for other income receipts. The main lines of influence are set out in Figure 6-2. The reasoning behind these lines of influence, which will be mirrored in the regression equations to come later, will probably be obvious to the reader, and therefore I shall not tax your attention with a discussion of them. One key supposition of the structure of assumptions may now be considered established theoretically: "There is substantial evidence that families decide whether to apply for or remain on welfare in an economically rational fashion." (Lyon, 1977). ---------- Figure 6-2 ---------- THE SAMPLE AND DATA The data come from the l976 Survey of Income and Education (SIE), discussed in the afternote to Chapter 5. The dependent variable, TRANS, includes the following welfare payments: public welfare, aid to dependent children, unemployment compensation, supplemental security, and food stamps. Again, it should be noted that such retirement allowances as Social Security and Medicare are excluded from the analysis; inclusion of them in any cost-benefit framework immediately indicates that immigrant families receive much less than do native families. Schooling also is excluded, which goes in the other direction than Social Security though it is less sizable. The number of children associated with a given family structure is the number at home aged 18 or under. Data on total children in the family were not collected by the Survey of Income and Education. RESULTS1 The results for this study are unusually difficult to communicate because of the large number of family-structure subsamples that must be examined, and because it is not obvious which group(s) should be compared with which. Table 6A-1 in the Afternote summarizes the main regression runs. Because of the large volume of data it contains, however, that table is really a reference document rather than a presentation of the main results. I shall attempt to bring out the main results in a few short tables drawn from the most general analysis presented in that table, the regression in column 1. That regression, which is derived from the entire sample of natives and immigrants, male- headed and female-headed, with various numbers of children or none, examines the effects of all the relevant variables, and includes family earnings on the assumption that the flow of reverse causation is not important. It's results, mainly that family structure has a dominant effect, are reproduced here as follows: Basic regression 1 results: Dollars received = $3189 (the amount received if the base case--female, single, 3+ children, low education, young, native born): less $1566 if female, single, two children; or less $1885 if female, single, one child; or less $2403 if female, single, no children; or less $2057 if male, married, 3+ children; or less $2192 if male, married, 2 children; or less $2163 if male, married, l child; or less $2222 if male, married, no children; or less $2316 if male, single, no children; or less $166 (if immigrant, 1970-74); and plus $60 (if immigrant, 1965- 69); and plus $54 (if immigrant, 1960-64); and plus $13 (if immigrant, 1950-59); and less $158 (if immigrant, 1920-49); and plus $23 (per year of family head's age); and less $31 (head's age squared); and less $43 (per year of family head's schooling); and less 0.02 (family earnings); and plus $133 (if English is poor). The same information is presented in different form in Table 6-2. Transfer payments received by families with various other structures are compared to the base-case family that includes a native young single female head and three or more children, the mother having minimum education and zero earnings. That base- case family collects much more welfare--on average $3169--than do male-headed families and families with fewer children. An otherwise-similar female-headed family with only two children gets $1623, or $1566 less than the base case family. And so on. The range among all families other than female-headed with two or more children is only $431. Regressions with other sub-samples show that these amounts remain much the same no matter how the analysis is conducted. (The absolute amounts received by the various groups are also shown in Table 6-2, but it should be remembered that these amounts are for families with the poorest earning-power characteristics and the lowest earnings. --------- Table 6-2 --------- Turning to the central interest of this book, Table 6-2 shows that the effect of being an immigrant is quite small and inconsistent, as long as English ability is held constant; this may also be seen in all the other regressions with the total sample and sub-samples. Poor English increases welfare receipts by $133. The amounts received apparently are smaller for immigrants than for natives for the first five years, then rise to the native level after the first five years, and then are again smaller than natives among those in the country the longest. However, the variation in results among the various runs suggests that the small observed variation is due to sampling error; from a practical point of view it does not matter much one way or the other, however. Welfare receipts has an elasticity of -.46 with respect to family earnings, indicating that a halving of earnings almost doubles transfer payments received, ceteris paribus. And the elasticity with respect to schooling is -.6l; halving schooling more than doubles transfers received, ceteris paribus. Another way of looking at the results of the same regression (Column 1 in Table 6A-1) with respect to the issue of immigrant effects is by comparison of various immigrant groups with the "average" native family. These results, shown in Table 6-3, make clear that once other variables are held constant (especially family structure) the fact of being an immigrant rather than a native is of no consequence whatever from the point of welfare receipts. --------- Table 6-3 --------- The overall picture, then, suggests that family structure is by far the most important factor, and the factors most directly related to being an immigrant--time in the U.S., and English ability--are not of great importance compared to family structure and perhaps earnings and schooling. The next step is to check further into this matter. Because English ability is closely related to being an immigrant, a regression was run that is similar except for leaving out the variables for English ability. The average amounts received by each immigrant cohort (except for the oldest) are higher than when English ability is not held constant, but the pattern still does not indicate that immigrants receive more welfare transfers than natives, ceteris paribus. (Results in column 2 in Table 6A-2 in the Afternote). The entry-cohort effects are not meaningfully different for the two sex sub-groups. But the fact that in both groups the earliest and latest entrants are relatively low in transfers received suggests that the observed effect is not due to sampling variation. In the subsamples where English ability is held constant, being an immigrant has no observable effect. And even English ability does not have a strong enough effect to be consistent across the various categories, given the size of our samples. Because family structure is so important a determinant of transfers, we want to know how immigrants and natives differ in this respect. Table 6-4 shows the distributions of families by family structure, for natives and for the various immigrant categories, in various age classes. (The variable for children indicates only the number of children under 18 at home.) By inspection of the data for those who came in l970-l974 and for natives, we see that in their early years in the United States-- which are the most important years with respect to policy decisions about admitting immigrants, because of the time preference for resources--among immigrants there are fewer female-headed families with several children, and for this reason (along with others) the welfare burden (even aside from Social Security and other transfers to the aged) is less for this group of immigrants than for natives. We also see that, among natives (by comparison of the 18-44 and 18+ columns) the category of single females without children is heavily dominated by older women who are likely to be relatively heavy users of welfare, rather than being young women who are relatively light users of welfare. (And of course in the background is the fact that the younger are more likely than are the older to be paying taxes for others' transfers.) After immigrants are longer in the U.S., family structure becomes more similar to natives of the same age. ----------- Table 6-4 ----------- Here it is important to recall that in any overall accounting of the effect of immigrants, age structure is all-important. Because immigrants come when young, they do not use much Social Security and Medicare, which are far and away the most important cause of welfare transfers in dollar terms. Therefore, immigrant families use much less total welfare services than do natives, on average, even including schooling; this is analyzed in detail in Chapter 4. RULE-OF-THUMB DECISION ANALYSES The full analysis above, if done soundly, provides the basis for a guide to selection of immigrants with lowest predicted use of transfer payments and social services (to be combined with a similar analysis of highest predicted earnings), if the society decided that this were to be an appropriate policy. But decision-makers may decide that it is desirable to use some characteristics and not others as criteria for admission. That is, it could conceivably make sense to allow (among non-family- completion and other special-category immigrants) only male professionals below age 40 from Hong Kong, until all applicants have been taken, then only males with l6+ years of schooling below age 40 from Italy, and so on. But decision-makers are likely to decide that discrimination by sex or ethnicity is unacceptable, and they may therefore ask about the predictions for only "acceptable" criteria such as education and age,2 though recognizing that this might result in implicit differentiation by ethnicity, sex, etc. We cannot know in advance what variables might be called for. Education, however, seems the likeliest, and therefore we shall make analyses for that variable by itself. And because the timing of transfer payments is very important-- payments in the early years in the country must sensibly be weighted much more heavily than payments in the later years--the analysis must take account of entry date. A flexible and informative format is to compare separate groups defined by amount of schooling and date of entry (or native status). Table 6-5 shows the transfers received by such groups. It is obvious that by admitting persons of the three highest education levels--or even more so, just the two highest- education levels--the resulting immigrants would receive much less welfare services than average for natives. In fact, by comparing the mean receipt of $554 per family for all families to the amounts received in l975 by, for example, the two highest- education groups in the l970-l974 $l80 and $l27, respectively, cohort which are $364 and $4l7 less than the average of natives, we see that the receipt of transfers by these immigrant groups is very low by any absolute as well as relative standards. --------- Table 6-5 --------- Rules based on family structure would not seem feasible because of the difficulty of predicting at entry what future family structure will be. Therefore, they will not be considered here. SUMMARY AND CONCLUSIONS Because of their size, transfer payments now have great influence on the U.S. economy and on public budgets. Retirement allowances are most important. Because immigrants usually come when they are young, and because they are not entitled to benefits, they contribute to Social Security funding but do not draw benefits for many years (at which time their children are paying in to support them). Immigrants are obviously desirable to natives from that point of view. But other welfare payments might be more heavily used by immigrants than by natives. The aim of this chapter is to analyze the determinants of welfare payment use, with special attention to the role that being an immigrant plays. Family structure--and especially the presence of female- headed families with two or more children--is the most important determinant of the rate of use of transfer payments (aside from Social Security, Medicare, and other retirement transfers). Immigrant families are represented proportionally in the high-use categories. Earnings and schooling also influence welfare receipt. Immigrants do not differ greatly from natives on average in these categories, so there is no important differential effect on welfare receipt. The fact of being an immigrant has little or no effect on welfare receipt when the other relevant variables are held constant. Differential admission rules based on occupation or education could result in immigrants receiving much less welfare than natives, and almost none in absolute amounts. Such rules could be effective agents of discrimination from the point of view of native income. 86-82 Transf5 1/15/87 FOOTNOTES 0.1From the same SIE data set used here, Blau (1984) estimated the probabilities of families receiving public assistance, in contrast to the amounts estimated here. She began her work in conjunction with the analyses reported here. The two projects were separated at my suggestion when we did not agree about the usefulness of the two modes of estimation. 1Additional detail may be found in the longer technical paper on this subject to be published in the journal literature (Simon, no date). 2The presentation of this material does not imply that the writer endorses any such plans; under most conditions, he probably would not. Rules based on occupation, which would lead to similar results, would seem to have fewer objections. 86-82 Transf5 1/15/87 REFERENCESFrancine E. Blau, "Immigration and Labor Earnings in Early Twentieth Century America," _R_e_s_e_a_r_c_h_ _i_n_ _P_o_p_u_l_a_t_i_o_n_ _E_c_o_n_o_m_i_c_s, Volume 2, ed. Julian L. Simon and Julie Da Vanzo (Greenwich, Conn.: JAI Press, Inc., l980), 2l-4l. Barry R. Chiswick, "The Effects of Americanization on the Earnings of Foreign-Born Men," _J_o_u_r_n_a_l_ _o_f_ _P_o_l_i_t_i_c_a_l_ _E_c_o_n_o_m_y, Vol. 86, No. 5 (October l978), pp. 897-92l. Daniel S. Hamermesh, "Entitlement Effects, Unemployment Insurance, and Employment Decisions," _E_c_o_n_o_m_i_c_ _I_n_q_u_i_r_y, Vol. l7, July, l979, 3l7-32. Julian L. Simon, "What Immigrants Take From, and Give To, the Public Coffers," in _U_._S_._ _I_m_m_i_g_r_a_t_i_o_n_ _P_o_l_i_c_y_ _a_n_d_ _t_h_e_ _N_a_t_i_o_n_a_l_ _I_n_t_e_r_e_s_t, Staff Report of the Select Commission on Immigration and Refugee Policy, Appendix D (Washington: GPO, l98l). Julian L. Simon and Richard Sullivan, "A Replication and Extension of Chiswick's Estimate of the Earnings Pattern of Immigrants," forth- coming. D/l3 86-82 Transf5 1/15/87 TABLE 6-2 EFFECT OF FAMILY STRUCTURE UPON WELFARE RECEIPTS (Excludes Social Security) Amount Less Than Receives Base Case Base case: Single mother, 3+ children, low education, young, low earnings, native born. $3189 N.A. $3189 ___________________________________________________________________________ __________________________________________________________________________ Instead, native mother, 2 children $1623 $1566 " , native mother, l child 1304 1885 " , native mother, 0 children 786 2403 " , married natives, 3+ children 1132 2057 " , married natives, 2 children 997 2192 " , married natives, l child 1026 2163 " , married natives, 0 children 967 2222Instead, otherwise same as base case, immigrant arrived 1970-1974 3023 166 " " " 1965-1969 3189 0 " " " 1960-1964 3243 - 54 " " " 1950-1959 3202 - 13 " " " 1920-1949 3189 0 " Poor English 3322 - 133 " additional year of age 23 -------------------------------------------------------------------------- Source: Column 1 in Afternote Table 6-A1. TABLE 6-3 EFFECT OF IMMIGRATION UPON WELFARE RECEIPTS (ex. S.S.) Average family receipts in transfers (ex. S.S.) each year $ 514 ___________________________________________________________________________ ___________________________________________________________________________ If head an immigrant who arrived 1970-1974, otherwise- average family, receipts $ 348 Same but 1965-1969 574 Same but 1960-1964 568 Same but 1950-1959 528 Same but 1920-1949 356 ___________________________________________________________________________ Source: Column 1 in Afternote Table 6A-1 TABLE 6-5 Average Value of Welfare Transfers by Years of Entry and SchoolingYears of Years of School_ _E_n_t_r_y_ _ _ _ _ _ _ _0_ _ _ _ _ _ _ _l_-_4_ _ _ _ _ _ _5_-_8_ _ _ _ _ _9_-_l_l_ _ _ _ _ _ _l_2_ _ _ _ _ _l_3_-_l_5_ _ _ _ _ _ _l_6_ _ _ _ _ _ _l_7_+_ _ l970-74 $632 $933 $955 $868 $650 $337 $l54 $95 l965-69 l443 ll57 l299 ll87 656 376 329 2l5 l960-64 8l9 l659 885 l0l9 637 456 77 l44 l950-59 809 ll22 822 l055 547 448 33l l70 l920-49 601 4l3 349 470 306 4ll 228 l99 NATIVES l273 852 6l3 747 455 43l 237 l66 _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ Average Value for total sample = $500