Age at Immigration and the Incomes of Older Immigrants, 1994–2010

Abstract

Objectives.

Seniors comprise a growing proportion of new U.S. immigrants. We investigate whether late-age immigrants are disadvantaged in older age relative to those arriving earlier in life, based on income, reliance on public benefits, and access to public medical insurance. We test whether the 1996 welfare reform law altered the relationships between age at immigration and these outcomes.

Method.

Immigrants aged 65 and older in the 1994–2010 Current Population Surveys were classified by age at immigration. Median and logistic regressions are used to estimate the association between age at immigration and several outcomes and to test whether these associations differ for arrivals before and after welfare reform.

Results.

Late-age immigration is strongly associated with lower personal income, lower rates of Medicare and Social Security receipt, and higher participation in Supplemental Security Income (SSI) and Medicaid. Arrival after 1996 is associated with lower rates of SSI, Medicaid, and Medicare receipt. The association between late-age immigration and income is stronger for post-1996 arrivals relative to earlier arrivals, whereas that between late-age immigration and Medicaid is weaker, suggesting that the penalty conferred by late-age immigration grew after reform.

Discussion.

Late-age immigrants face formidable economic disadvantages exacerbated by exclusion from public benefits, with implications for immigration, health care, and welfare policy.

About 12% of the 40.4 million seniors residing in the United States are foreign born (Batalova, 2012). Most migrated early in life, but a substantial and growing number of immigrants arrive in the United States at advanced ages or late in their working lives (Carr & Tienda, 2013). This trend partly results from public policy choices: the 1965 Amendments to the Immigration and Nationality Act added parents of U.S. citizens to the family reunification priority categories that are exempt from worldwide and country-specific caps on the number of immigrants admitted. That same year, the U.S. Congress created the Medicare and Medicaid health insurance programs by amending the Social Security Act. Neither soaring health care costs nor population aging was a salient concern for lawmakers in the mid-1960s; but as the baby boom generation begins retiring and health care costs continue to soar, both issues have implications for the future welfare costs of all seniors, and foreign-born seniors in particular. Against the backdrop of an outsized national debt and high health care costs, calls for comprehensive immigration reform may draw attention to the fiscal and social implications of criteria for sponsoring relatives.

Unlike Australia, Canada, and some other countries of immigration, the United States does not consider age when determining eligibility for admission. Despite historical concerns that immigrants will become public charges, U.S. admissions policies contain only limited measures to ensure that immigrants will be self-sufficient. For example, personal assets are not considered in determining eligibility for most permanent resident visa categories, and the low-income threshold required to sponsor relatives—125% of the poverty line based on a family size that includes the new immigrant—provides a weak guarantee for self-sufficiency. In addition, late-age immigrants may be highly susceptible to poverty and economic dependency both because the requirements intended to ensure that new immigrants rely on the resources of sponsoring family members rather than public assistance are not systematically enforced and because many seniors may find it difficult to build private income and assets during what remains of their working lives (Angel, Angel, Lee, & Markides, 1999; Borjas, 2009) or to qualify for public pension and medical insurance programs linked to employment (Borjas, 2011; Nam, 2012).

Federal public welfare policy is largely indifferent to the potential economic challenges faced by late-age immigrants, which several analysts claim were exacerbated by the cost-cutting measures enacted in the mid-1990s (Angel, 2003; Van Hook, 2003). Partially prompted by a rise in applications for federal benefits among seniors and refugees (Van Hook & Bean, 1999), sweeping welfare reform legislation passed in 1996 curtailed new immigrants’ access to several means-tested welfare benefits for at least 5 years after admission (Borjas, 2003; Nam, 2008, 2012; Nam & Jung, 2008). Furthermore, the 1996 Illegal Immigration Reform and Immigrant Responsibility Act (IIRIRA) required new sponsors of family immigrants to attest that they would maintain the new immigrant above the poverty line until they acquired either citizenship or 40 quarters of qualified work experience (U.S. Department of Health and Human Services, 2009; Vialet, 1997). Although these changes affected immigrants of all ages, they may be particularly consequential for older immigrants, whose earning prospects are generally lesser and health needs greater compared with prime-age workers. Importantly, because the bans on access to means-tested benefits are time limited, late-age immigrants may subsequently place greater demands on means-tested public insurance programs after the 5-year ban expires (Nam & Jung, 2008).

Several studies have examined the impact of welfare reforms on program participation of immigrants (Binstock & Jean-Baptiste 1999; Fix & Tumlin, 1997; Van Hook, 2003; Van Hook & Bean, 1999); however, with few exceptions (Borjas, 2011; Nam, 2008, 2012; Nam & Jung, 2008), the economic circumstances of late-age immigrants in older age have been largely ignored. The combination of an age-blind immigrant admissions policy, relatively restricted access to social welfare programs, and minimal income thresholds for immigrant sponsors has two important hypothetical implications for understanding the association between age at immigration (AAI) and economic well-being of foreign-born seniors: One is that late-age immigrants face worse income prospects as seniors than immigrants who arrived during their prime working ages; a second is that at ages 65 and over, late-age immigrants will use means-tested health and welfare benefits at higher rates than foreign-born seniors who arrived during their prime working ages.

Compared with immigrants who arrived at younger ages, therefore, we expect late-age immigrants will have lower incomes, higher rates of Supplemental Security Income (SSI) and Medicaid (both of which are means tested) receipt, and lower rates of Social Security and Medicare receipt during their golden years. Furthermore, because of the new restrictions on access to means-tested benefits after welfare reforms, we expect (a) that post-1996 immigrants of all ages use SSI and Medicaid at lower rates than pre-1996 arrivals, and (b) for immigrants arriving after 1996, there will be much smaller differences in use of SSI and Medicaid across AAI groups (relative to those arriving before 1996) as the more stringent eligibility requirements curb the ability of those arriving in older age to access benefits.

Overall, we find that there is a large income penalty associated with immigrating in older age, and that the relative disadvantage of arriving in older age appears to be greater for immigrants who arrived after welfare reform. Seniors who immigrated at advanced age are more likely to use means-tested benefits and less likely to receive entitlement benefits, relative to those who immigrate while young. We show that arrival after 1996 is associated with lower probability of receiving means-tested benefits (SSI and Medicaid) and Medicare, and that in the case of Medicaid lower participation was concentrated among seniors who immigrated at the most advanced ages. Thus, the disadvantages of late-age immigration as experienced at age 65 and beyond appear to be greater for people who arrived during in the “new age” of U.S. welfare policy.

Background

In 2010, the median age of the U.S. foreign-born population was 41 years compared with 36 years for the native-born population (Grieco et al., 2012). Immigrants’ higher median age reflects the higher proportion of children in the native population (26% vs. 7%). Considering only the adult population, the foreign-born population has a larger proportion under age 45 compared with the native population (54% vs. 48%) and a smaller share ages 65 or greater (13% vs. 18%; authors’ calculations based on Grieco et al. 2012). While relatively small, the number of foreign-born seniors has grown over time. He (2002) reports that between 1960 and 1990, the foreign-born population aged 65 and older was relatively stable, hovering around three million. Since 1990, however, the number of foreign-born seniors nearly doubled, rising from 2.7 million to almost 5 million in 2010 (Batalova, 2012).

Two mechanisms account for the aging of the foreign-born population: aging in situ of immigrants who arrived during their youth or prime working ages and immigration at older ages. The former is the primary driver of growth of the elderly immigrant population, but late-age immigration has been rising as well. This distinction, moreover, is important for understanding both the economic well-being of foreign-born seniors and the welfare costs of immigration. Based on the American Community Survey, Terrazas (2009) estimates that the number of retirement-age seniors admitted as legal permanent residents (LPRs) nearly doubled between 1999 and 2008 but stabilized thereafter. Carr and Tienda (2013) use administrative data to show that the cohort share of LPRs who were aged 50 and older at admission rose from about 11% for LPRs admitted between 1981 and 1985 to nearly 17% of legal permanent admissions between 2006 and 2009, mostly due to sponsorship of parents by U.S. citizens.

Despite the growing numerical significance of late-life immigration, most existing studies of foreign-born seniors’ economic well-being and social program participation ignore AAI, focusing instead on comparisons between the native- and foreign-born populations, with some attention to variations by immigration cohort, length of U.S. residence and/or naturalization status, but not AAI per se (Borjas, 2009; Gerst, 2009; Ku, 2009; Nam, 2008; Van Hook, 2000). A large body of research finds large positive associations between length of U.S. residence and myriad indicators of integration and well-being such as income, English proficiency, and naturalization (Espenshade & Fu, 1997; Van Hook, 2003; Vigdor, 2009). Late-age immigrants presumably have less time to reap these gains, and to accumulate financial and other assets before old age, compared with seniors who arrive at relatively youthful ages (Borjas, 2011; Treas & Mazumdar, 2002). AAI also has implications for the health needs of immigrants because prime-age workers generally require fewer health care services than seniors, other things equal (Zallman, Woolhandler, Himmelstein, Bor, & McCormick, 2013).

With a few recent exceptions (Angel et al., 1999; Borjas, 2011; Nam, 2008, 2012; Nam & Jung, 2008), there is limited empirical evidence directly evaluating the economic and welfare consequences of late-age immigration. Espenshade and Fu (1997) report that individuals who migrate later in life are less proficient in English, potentially limiting their labor market options. Angel and Angel (2006) explain that late-age immigrants not only experience high distress at relocation and hence incur greater service needs, but also arrive with little to no retirement savings. Later life immigration also is associated with lower rates of health insurance coverage (Angel, 2003), especially after the 1996 welfare and immigration reforms (Borjas, 2003; Nam, 2008; Van Hook, 2003). Angel et al. (1999) show that prior to welfare reform, for Mexican immigrants residing in Southwestern states, late-age immigration was associated with low personal income, high dependence on family support, and low use of means-tested and entitlement programs. Although their inferences are consistent with national studies, because the study is based on a small regional sample, its external validity is limited.

The intersection between AAI and public policies also puts late-age immigrants at an economic disadvantage both because many do not acquire the 10 years of covered employment needed to qualify for Social Security and Medicare benefits (Borjas, 2011; Sevak & Schmidt, 2007) and because immigration and welfare reforms that restricted access to several means-tested benefits drew sharper lines between noncitizens and citizens (Gerst, 2009; Gerst & Burr, 2009; Van Hook, 2003). Immigrants who pay Old Age, Survivors, and Disability Insurance (OASDI) taxes on wages for 40 or more quarters (10 years) are eligible to receive both Social Security and Medicare, the U.S. public health insurance program for older adults, at no cost. After 5 years of U.S. residence or upon obtaining citizenship, otherwise eligible seniors may also buy Medicare coverage (both Part A and Part B) in lieu of qualifying through the work requirement (Centers for Medicare and Medicaid Services, 2013).

Late-age immigrants are presumably less likely than natives to qualify for Social Security and free Medicare in older age. This is partly because late-age arrivals have a shorter remaining working life in which to accumulate U.S. work experience and partly because they lack the English and other skills needed to obtain steady employment (Angel et al., 1999; Borjas, 2011; Espenshade & Fu, 1997). The relationship between AAI and work experience is complicated, however. Borjas (2011) claims that the 10-year work rule is a powerful incentive for older immigrants to remain in the labor force at older ages in order to qualify for Social Security. Using census data from 1960 to 2000 to assess nativity differentials in Social Security eligibility, he estimates that about 20%–30% of immigrants over the age of 62 did not qualify for Social Security because they had not accumulated the requisite 40 quarters of qualified employment. Although his analysis did not consider the period after 2000, he acknowledges that the ineligibility rate rose over time (p. 492). Because the share of late-age immigration also increased (Carr & Tienda, 2013), the number of vulnerable seniors has likely grown as well, especially during the slack markets of the early and late 2000s.

Means-tested public welfare benefits provide a nonmarket income source for seniors, including the foreign-born. Two of the most important programs for low-income seniors are Medicaid and SSI. As a federally funded, state administered health insurance program for individuals with limited income and assets, Medicaid does not require recipients to qualify through work and can supplement Medicare for individuals who qualify for both programs. Medicaid was available to qualifying LPRs upon arrival, until the 1996 welfare reform. SSI provides cash income to poor people who are blind, disabled, or above age 65 with limited income and assets (Daly & Burkhauser, 2003). Unlike Social Security, there was no employment requirement for receiving SSI benefits, and immigrant seniors could apply for benefits after only 3 years of U.S. residence (until 1996). This provoked charges that SSI had become magnet for late-age immigrants (see Dunn, 1995; Rector, 1996). Van Hook and Bean (1999) show that the growth in SSI caseloads between 1979 and 1996 was largely due to rising usage rates among settled residents (50% of the caseload growth); however, they acknowledge that the growth of the late-age immigrant population accounted for one third of the caseload growth before 1997, making older immigrants an easy target for budget cuts.

In 1996, Congress passed the Personal Responsibility and Work Opportunity Reconciliation Act (PRWORA), a major welfare reform that included provisions to promote self-sufficiency among immigrants and to remove purported financial incentives for migrating to the United States (Binstock & Jean-Baptiste, 1999; Broder, Wheeler, & Bernstein, 2005). PRWORA made immigrants with less than 5 years of U.S. residence ineligible for Medicaid. It also restricted SSI benefits to U.S. citizens, but these benefits were restored to immigrants who had been receiving benefits before August 1996; the Balanced Budget Act of 1997 also made it easier for immigrants whose benefits were restored to qualify for Medicaid (Binstock & Jean-Baptiste, 1999, p. 32). Three classes of LPRs also retained access to Medicaid and SSI: those who both qualified for Social Security and either had 5 years of LPR status or lived in states that link Medicaid to SSI eligibility; foreign-born U.S. military veterans; and for a limited time after their admission, qualified refugees and asylees (National Law Immigration Center, 2002, p. 121). PRWORA also required that states include the income of the immigrant’s sponsoring family member when determining eligibility for SSI, Medicaid, and other means-tested benefits until they acquired citizenship (Binstock & Jean-Baptiste, 1999). One effect of all these changes, especially in eligibility for SSI, was to implicitly increase the value of citizenship because immigrants would upon naturalization attain the same rights to welfare benefits as U.S. natives (Nam, 2012; Nam & Jung, 2008; Van Hook, 2003).

States retained some ability to compensate for PRWORA’s cuts. A minority of states used their own funds to extend Medicaid benefits to LPRs during the 5-year ban. Five states created substitute income-support programs for some groups of immigrants that PRWORA rendered ineligible for SSI; however, most of these benefits were less generous than the federal program or subject to more stringent conditions relative to SSI in the pre-1996 period (National Immigration Law Center, 2002; Zimmerman & Tumlin, 1999).

The Patient Protection and Affordable Care Act of 2010 adds another dimension of complexity because legally resident immigrants are eligible for subsidized private health insurance via tax credits under that law and are subject to the individual insurance mandate (National Immigration Law Center, 2013). Nevertheless, the federal Medicaid bar remains, which means that the recent healthcare reform law is unlikely to improve access to health insurance for the poorest, most recent immigrants.

These reforms spawned a spate of research evaluating their impacts on the program participation and economic well-being of various segments of the foreign-born population (Ku, 2009; Ku & Kessler, 1997; Van Hook, 2003). Studies focused on seniors examine either transfer programs or health insurance, concentrate on short-term effects (i.e., through 2000 or shortly thereafter), and give limited consideration to AAI. Gerst (2009) shows that the proportion of foreign-born Latino seniors who were barred from SSI by their immigration status actually dropped between 1990 and 2000, even among those who had arrived within 10 years. This is likely due to both increased rates of naturalization in response to the new restrictions (protective citizenship) and the restoration of eligibility to pre-1996 arrivals (Van Hook, 2003). Gerst and Burr (2011) find that after 1996, the rates of SSI receipt for eligible elderly Latino noncitizens declined more than those for citizens but were not affected by state-level welfare generosity.

For older immigrants, several recent studies evaluated two key hypotheses about the consequences of welfare reform for immigrants in general—namely, that there was a “chilling effect” dissuading otherwise eligible immigrants and their family members from using welfare benefits, and that a spike in naturalization rates protected access to benefits for longer term residents after 1996 (Van Hook, 2003). Nam’s (2008) analysis of the 1994–1996 and 2003–2005 Current Population Surveys (CPS) provides support for both the “chilling effect” and “protective citizenship” hypotheses. Specifically, she finds significant differences in Medicaid coverage by citizenship status, with lower rates among noncitizens after the welfare reforms (chilling effect) and higher rates among naturalized seniors (protective citizenship). Using 1999 CPS data to examine food stamp participation among older immigrants, Nam and Jung (2008) show that state generosity in eligibility criteria benefitted noncitizen seniors indirectly, namely through access of eligible household members. Not all transfer programs consider household composition when determining eligibility, however; Medicaid, Medicare, and SSI are important examples where eligibility does not depend on living arrangements (although for SSI the benefit level does depend on household structure).

In a more recent study, Nam (2012) uses a triple difference-in-difference approach to examine variations in seniors’ Medicaid and health insurance coverage by citizenship status before and after the 1996 reforms. Using CPS surveys for the period before (1994–1996) and after (2003–2008) welfare reforms combined with newly released state Medicaid eligibility data, she finds that Medicaid coverage declined among noncitizen seniors but increased among naturalized citizens; moreover, changes in coverage of foreign-born seniors differed according to whether states supplemented shortfalls in federal Medicaid funding for noncitizens. Unlike other studies, Nam (2012) considers immigrants’ program participation beyond 2000; however, like all prior studies, she does not consider differences by AAI. As argued above, AAI should influence the likelihood of qualifying for SSI, Medicare, and Social Security either through naturalization (SSI) or through acquiring the requisite labor force credits (Social Security and Medicare).

The current immigration and welfare policy context, along with substantial indirect evidence that greater AAI confers disadvantages in retirement, raises three testable propositions. First, we hypothesize that late-age immigrants will have lower incomes in old age compared with immigrants who arrived at younger ages, and that they will be less likely to receive Social Security and Medicare. This hypothesis is implied in studies establishing an association between length of U.S. residence and self-sufficiency; however, the existence and size of the association between AAI and retirement economic well-being have not been established for the foreign-born population. This leaves open the possibility that late-age immigrants are able to offset these disadvantages by working to later ages (Borjas, 2003, 2011), by arriving with greater assets, or through other means, such as higher rates of means-tested benefit use (SSI and Medicaid). If late-age immigrants use SSI at higher rates than younger arrivals, then age variation in personal income levels will be greater when SSI income is excluded.

Second, as already established by others (Angel, 2003; Nam, 2005, 2008; Van Hook, 2000, 2003; ), we hypothesize that foreign-born seniors who arrived after the 1996 cutoff use means-tested public benefits (SSI and Medicaid) at lower rates, irrespective of AAI; however, we do not expect changes in use of entitlement programs whose eligibility criteria were unaffected by welfare reform.

Finally, we expect that the associations between AAI and indicators of economic well-being changed for immigrants who arrived before and after 1996. Specifically, we hypothesize that PRWORA produced two subtly different, but mutually compatible effects on the AAI–income relationship. On one hand, we hypothesize that welfare reform weakened the positive association between AAI and use of the public benefits that were targeted by the 1996 law (SSI and Medicaid) while leaving the relationship between AAI and other sources of support (personal income excluding SSI, Social Security, Medicare) in older age unchanged. PRWORA made time in the United States a requirement for accessing SSI and Medicaid (potentially via naturalization), which should impede those entering at older ages from participating in these programs compared with immigrants who arrived at younger ages. On the other hand, PRWORA may have discouraged SSI and Medicaid participation among immigrants of all arrival ages, potentially affecting all AAI categories uniformly. Both sides of this hypothesis imply that arriving at an older age will confer even greater overall income disadvantages for immigrants who arrived after 1996. Thus, late-age immigrants will have borne the brunt of the changes predicted by the second hypothesis.

Data and Methods

To test these hypotheses, we pool data from the March annual demographic supplements to the CPS between 1994 and 2010 to represent the period before and after the sweeping 1996 welfare reforms. Unlike most studies that evaluated the consequences of welfare reform for immigrants, our data span the first decade of the 21st century that includes the Great Recession (Nam 2012 is an exception). The long history of the CPS allows us to conduct analyses over a full range of AAI by age-at-survey combinations for persons entering both before and after the 1996 welfare and immigration reforms, making it preferable to alternative data sources, such as the American Community Survey. To ensure that respondents appear only once in the analysis, we restrict the sample to respondents who are in their first year of participation in the CPS. Another advantage of the CPS is the imputation of missing values for income and other key variables.

We consider only foreign-born respondents aged 65 and older at the time of survey. The full analysis sample consists of 16,902 observations, yielding at least 110 observations (distributed across various survey years) in each survey age by AAI cell resulting from the array of categorical variables. Both summary statistics and regression estimates are weighted using CPS-provided survey weights.

Dependent Variables

We examine several measures of economic well-being and program participation, including annual total personal income, annual total personal income excluding SSI, and indicators for receipt of Medicare and Medicaid. Personal income measures respondents’ self-reported income from all sources (including private pension and any other cash welfare benefits, but not in-kind benefits such as health insurance) adjusted to 1994 dollars. Medicare and Medicaid variables for years 2001 and later were made consistent with previous years by recoding (as uncovered) respondents whose coverage was captured only in a verification question introduced in 2001 (State Health Access Data Assistance Center, 2006). This adjustment reduces the estimated number of respondents in our post-2000 sample covered by Medicaid by about 0.6% and those covered by Medicare by about 0.5%.

Predictor Variables

We classify respondents both according to age at survey (65–69; 70–74; 75–79; 80+) and AAI (0–34; 35–44; 45–49; 50–54; 55–59; 60–64; 65+). The latter distinguishes between immigrants arriving during prime working ages and those arriving at later ages, with greater detail for late-age immigrants who arrived after age 45. Two aspects of the AAI variables warrant further discussion. First, the CPS asks foreign-born respondents “In what year did you first come to the United States to stay?” The wording of the question raises the possibility that respondents may have entered the United States prior to coming “to stay.” It also means that the respondent could have received LPR status either before or after coming “to stay.” Second, the CPS reports year of entry as a categorical variable, further reducing measurement precision. For the entry years closest to the survey year, the arrival year categories consist of two- or three-year bands, but these bands increase to five years for immigrants who entered before 1980 and ten years for pre-1959 arrivals. The earliest arrival cohorts are designated “before 1950.” For 56 percent of respondents, age of arrival “to stay” can be established within five years or less, and for 76 percent, within ten years or less. Precision is, fortunately, highest in the AAI groups of greatest interest (those arriving at advanced ages) because the bands are smaller for the most recent years. We calculate AAI using the midpoints of the bands in the CPS-provided categorical variables. For those falling into the “before 1950” category, we use the midpoint between 1950 and the respondent’s year of birth.

The CPS top-codes respondents’ age at the time of the survey at 90 (until 2002), 80 (2002 and 2003), or 85 years of age (2004–2010). This further complicates the calculation of the AAI variable for top-coded respondents, who represent 9.3% of the analysis sample. For the purposes of calculating AAI, we treated the top-coded survey ages of 80, 85, and 90 as 85, 89, and 93, respectively. This approximation is based on the mean ages of immigrants falling above these thresholds in 2000 and 2010 Census microdata, which do not top-code age.

The variable describing period of immigration classifies respondents by welfare regime, that is, whether they were subject to the more stringent benefit eligibility rules imposed on immigrants who received their green card after August 22, 1996. Respondents who entered to stay after this date are classified as “post-1996”; those who entered to stay before this date are classified as “pre-1996.” A handful of respondents for whom we could not ascertain an arrival date unequivocally (due to the categorical year of immigration data) are classified as “transition.” Because the period of immigration classification is based on self-report data rather than official dates when respondents received LPR status (if they did so), possible measurement errors would tend to attenuate estimated differences between pre- and post-1996 immigrants.

Our analysis compares immigrants who entered the United States over a wide range of years that brought substantial changes in the region of origin and educational profile of the U.S. foreign-born population (Grieco et al., 2012). Therefore, in regressions we use indicator variables to control statistically for variations in level of educational attainment (less than high school, high school, and college) and regional origins (Asia, Africa, Latin America & Caribbean, Europe, or Anglophone developed nations), as well as age at survey and sex. We also use survey year indicators to control for year-to-year fluctuations in economic conditions. We do not control for other commonly used variables, such as English language proficiency, that we view as causal intermediaries between AAI and the outcomes of interest. We conducted additional robustness checks to determine whether results were sensitive to state-level variation in post-PRWORA welfare generosity toward immigrants during the 5-year ban on means-tested benefits using the Zimmerman and Tumlin (1999) classification of states as generous or not (Borjas, 2003; Nam, 2012). Results (available on request) are substantively identical to those reported subsequently.

Models

To address the first two hypotheses concerning how the life-cycle timing of migration is associated with economic well-being of foreign-born seniors and differences in economic well-being during old age for pre- and post-1996 arrivals, we estimate median (quantile) regressions for the two income variables and logistic regressions for indicators of Social Security, SSI, Medicare, and Medicaid receipt on indicators for AAI and period of immigration (relative to the 1996 Welfare Reform Law). All models include relevant control variables mentioned above. Median regressions were used for the income variables because they are less sensitive than OLS to extremely large income values, yet allow easy substantive interpretation of coefficients. Because standard statistical packages do not permit simultaneous use of weights and cluster-robust standard errors for median regressions, we calculate standard errors from nonrobust standard errors that were multiplied by a design factor. This design factor (which changes the values of the standard errors between −1% and 4.5%) is the square root of the ratio of the mean of the state cluster-robust standard variances to the mean of the nonrobust standard variances when the same model is run without survey weights. The design factor does not alter substantive results. For our logistic regressions, we present standard errors that are robust to clustering at the state level.

To test whether the association between AAI and various indicators of economic well-being in older age differs for immigrants who arrived before and after 1996, we introduce interaction terms between the post-1996 arrival indicator and the AAI categories. For these analyses, we exclude immigrants classified as “transition” arrivals—those we were unable to assign unequivocally to the pre- and postreform periods. Furthermore, because the younger entry cohorts from the post-1996 period have not yet reached age 65, these analyses are restricted to immigrants whose AAI was 55 or above.

Results

Table 1 presents summary statistics for the analytic sample of immigrants aged 65 and older, pooled for years 1994–2010. The majority of foreign-born seniors immigrated prior to age 45, which is consistent with a vast literature showing that young and working-age persons dominate U.S. immigration flows. However, about 29% of the sample immigrated to the United States at ages 50 or above.

There are also important differences in the noneconomic characteristics of foreign-born seniors according to their AAI. Importantly, late-age immigrants are less likely to be naturalized, which became a key differentiator in access to public benefits after 1996. Among seniors who migrated at or after age 60, Asians are overrepresented, whereas seniors from Latin America and the Caribbean are more highly represented among the younger AAI groups, consistent with patterns of family immigration observed by others (Carr & Tienda, 2013). Although the proportion of immigrants with a bachelor’s degree is relatively uniform across AAI categories, seniors lacking a high school degree represent a higher proportion of late-age compared with working-age immigrants.

 

Both summary statistics and the first round of regression analysis (Table 2) support our claim that AAI is strongly associated with personal income. The average immigrant entering under age 35 has a mean total personal income in older age of about $17,894 versus $6,330 for a senior who immigrated at age 65 or later (Table 1). Controlling for the regional and educational composition of the immigrant population does not alter this relationship: the predicted median income for seniors who immigrated between ages 55 and 59 was $3,473 less than that of seniors who were under 35 years at arrival; and for immigrants who arrived at age 65 or above, the estimated median income gap versus the under age 35 arrivals was $4,934. Also as hypothesized, SSI income partially compensated for the disadvantages associated with greater AAI: Once SSI income is excluded, the gaps in predicted median income across the younger AAI groups widens (Table 2, column 2).

This hypothesis finds additional support in the strong associations between odds of receiving SSI and AAI (Table 2, column 3), as well as in the summary statistics showing that, on average, SSI represents around 20% of mean personal income for immigrants who entered at ages 65 and above compared with less than 1% for seniors who arrived before age 35. The income gaps among seniors according to age at arrival are partly due to the difficulties late-age immigrants face in qualifying for Social Security: Immigrating at age 35 is associated with over six times higher odds of receiving Social Security relative to immigrating at ages 55–59, and 16 times higher odds relative to immigrating above age 65 (Table 2, column 4). Surprisingly, for seniors who immigrate after age 65, Social Security still represents about one third of personal income (Table 1). This likely reflects some combination of spousal survivor Social Security benefits, misreporting of income sources, benefits earned during residence in the United States prior to entering “to stay,” or benefits partially earned through credits from abroad. For example, totalization agreements, which also benefit some U.S.-native workers, allow workers to qualify for Social Security with as little as six quarters of U.S. work (see Migration Policy Institute, 2004; Social Security Administration, n.d.). Unfortunately, CPS data do not reveal how Social Security credits are earned.

Higher AAI is positively associated with receipt of Medicaid at advanced ages (Table 2, column 5). The pattern of associations between Medicaid participation and AAI is quite similar to that observed for SSI receipt. The associations between Medicare participation and AAI categories run in the opposite direction as those for Medicaid and are both statistically significant and substantively important (Table 2, column 6). Consistent with the summary statistics (Table 1), the estimated differences between AAI categories for Medicare are generally smaller than those for Social Security. This may appear surprising given the overlap in eligibility requirements between Social Security and “free” Medicare but might be partly explained by benefits gained through a qualifying spouse and by a little-known option that allows seniors who do not qualify for free Medicare Part A to pay for it.

 

As expected, arriving after the 1996 changes in eligibility for means-tested benefits imposed by PRWORA is associated with lower median incomes from all sources, including SSI (Table 2, Column 1). When SSI is excluded (Table 2, Column 2), there is a statistically significant difference in median incomes of seniors according to period of arrival with other variables controlled; however, this coefficient (−$124) is small compared with the same coefficient when SSI income is included (−$1,732). Furthermore, seniors who arrived after 1996 also had lower odds of receiving SSI (Table 2, Column 3) and Medicaid (Table 2, Column 5). As hypothesized, there is no association between immigrating after 1996 and receipt of Social Security, which was not targeted by either the IIRIRA or PRWORA legislation. Nevertheless, we find a significant, negative association between Medicare receipt and post-1996 immigration. This difference might be explained because older immigrants can obtain Medicare without qualifying for Social Security by paying premiums. In that case, the 5-year ban imposed in 1996 would thus reduce the number of immigrants able to purchase Medicare, but would not affect Social Security receipt, consistent with the analysis here. Despite the clear differences in income and welfare participation between foreign-born seniors who arrived before and after the reforms, it is not clear whether and how the restrictions on benefits differed according to AAI, which is the focus of the last set of analyses.

 

The regressions reported in Table 3 introduce an interaction between period of entry and the AAI variables to test whether the deleterious economic consequences associated with late-age immigration were exacerbated by the 1996 welfare reforms. Because the younger entry cohorts from the post-1996 period have not yet reached age 65, for these analyses we restrict the analytic sample to foreign-born seniors who arrived at ages 55 and over. Although the hypothesis test is perforce weakened because it is conducted across a limited range of ages at immigration, the results are nonetheless instructive.

These analyses reveal that relationship between AAI and total personal income (including SSI) differs between pre- and post-1996 immigrants (Table 3, column 1). The coefficient on the base (uninteracted) AAI variables show that before 1996, seniors who immigrated above age 65 experienced a median income shortfall of about $937 relative to foreign-born seniors who arrived in the United States between ages 55 and 59. Seniors who immigrated between ages 55 and 64 received about $374 less per year than their peers who entered at a younger age.

Post-1996 arrival is associated with a $1,181 income advantage. However, once we consider the interaction coefficients, only the group aged 55–59 was better off if they arrived after reform. The coefficients on the two variables representing an interaction between immigrating at age 65 or above or between ages 60 and 64 and immigrating after 1996 are negative and statistically significant, indicating that the income and means-tested benefit gaps among the AAI groups were even greater for postreform arrivals. Specifically, for seniors who arrived after 1996, the total estimated gap in predicted median income between arrival ages 65+ and ages 55–59 is approximately −$4759; this is vastly larger than the comparable AAI median income gap among seniors who immigrated before the sweeping reforms (−$937). The gap is also larger in the postreform period for seniors who immigrated at ages 60–64 relative to those arriving at 55–59, −$374 versus −$3,358, respectively. Stated differently, seniors who arrived after age 60 were on net worse off if they arrived after 1996 compared with their age-at-arrival counterparts even after the $1,181 income advantage associated with post-1996 arrival is considered; the opposite is true for those arriving at ages 55–59.

Excluding SSI from personal income (Table 3, column 2) diminishes the magnitude of the interactions between AAI and period of arrival although they remain statistically significant. Simply stated, the gaps in total income between late age and earlier age arrivals were greater for post-1996 arrivals compared with prereform arrivals. Most, but not all, of the greater relative disadvantage appears to derive from SSI; however, there is no evidence that the association between AAI and probability of receipt of SSI (Table 3, column 3) differs before and after welfare reform because coefficients for the interaction terms fail to attain statistical significance. Consistent with the results reported in Table 2, arrival after 1996 is associated with lower probability of receiving SSI, regardless of age at arrival.

Estimates for Medicaid (Table 3, column 4) receipt differ somewhat. In this model, the coefficient on the variable indicating arrival after 1996 is not statistically different from one. However, the coefficient on the interaction between immigrating at age 65 and arriving after 1996 is significantly below 1, indicating that the association between arrival over age 65 and Medicaid is weaker in the post-1996 group. Thus, even though older age at arrival is associated with a higher probability of Medicaid receipt, late-age immigrants were less likely to receive Medicaid if they arrived after 1996 compared with their age counterparts who arrived prior to welfare reform. Another implication of this pattern of results is that the gap in Medicaid receipt between pre- and post-1996 arrivals appears to exist for those who arrived above age 65, but not for those aged 55–59.

Results for Social Security (Table 3, Column 4) are substantively similar to those in the baseline regressions shown in Table 2 (Column 4): There is no significant association between Social Security receipt and immigration after 1996, nor are any of the coefficients on the interaction terms significantly different from zero. Likewise, we find no evidence that the relationship between AAI and Medicare differs for pre- and post-1996 arrivals. If the 5-year bar is indeed responsible for reducing Medicare coverage as suggested in the previous set of analyses, there is no evidence that it disproportionately affected those who immigrated at older ages.

These results leave us with several clear and some less clear interpretations. In the case of Medicaid, results from Table 3 are consistent with delays in obtaining benefits: Seniors who both immigrated at the oldest age (thus having had less time to qualify for benefits) and arrived after 1996 are less likely to obtain Medicaid than their arrival-age peers who arrived before 1996. No such difference between the pre- and post-1996 periods is obtained for seniors who immigrated at younger ages. By contrast, there is no clear evidence that a mere delay in obtaining benefits drives down SSI and Medicare receipt rates: Rather, older immigrants were less likely to receive SSI and Medicare if they arrived after 1996, but the relationship between AAI and SSI receipt was similar in both periods. At first blush, this appears to be inconsistent with evidence that total personal income drops more with AAI in the postreform relative to the prereform period and that most of this drop can be traced to SSI; however, this result may simply reflect the fact that seniors entering in the oldest age category received the largest median cash value of SSI and thus stood to lose the most.

Limitations

One limitation of the analysis is the lack of precision in the CPS regarding the year respondents immigrated to the United States, which introduces error in the AAI groups and the pre-and post-1996 entry cohorts. The latter required deleting observations for indeterminate arrival periods in the second stage of the analysis. Both measurement problems should bias our findings toward smaller differences between groups, attenuated regression coefficients, and null findings, relative to the ideal of classifying respondents by the precise date they received LPR status. Other studies that rely on the CPS to study the consequences of welfare reform on foreign-born seniors share similar problems. A second major limitation is that because the youngest immigration cohorts had not yet reached retirement ages, analyses of the economic consequences of late-age immigration for the pre- and post-1996 entry cohorts were restricted to a narrow range of arrival ages, which perforce precluded comparisons of the differences between those arriving at the youngest ages and the oldest ages between the pre- and postreform arrival cohorts.

Although region of origin strongly predicts income and program participation, we did not examine whether the relationship between AAI and income in later life differs across regions of origin. Cultural expectations, skills, physical distance, and legal and financial statuses of other family members influence the likelihood of late-age immigration as well as their living arrangements and economic circumstances in the United States. Therefore, the relationship between AAI and income likely varies by region and country of origin; this possibility merits a separate investigation because the analysis sample does not permit country-specific analyses.

Discussion

We find that foreign-born seniors aged 65 and older who arrived in the United States at older ages have considerably lower personal incomes and are less likely to benefit from entitlement programs (Social Security and Medicare) compared with their peers who arrived at younger ages. Means-tested benefits provide a cushion against these earned benefits: Late-age immigrants are more likely than young immigrants to receive SSI and Medicaid, and the income disadvantage associated with late-age immigration is significantly worse when SSI income is excluded from their income portfolio. These differences persist among immigrants of comparable education and regional origins. Moreover, U.S. immigrants who arrived after the 1996 reforms use means-tested social benefits and Medicare at lower rates compared with their counterparts who arrived before 1996. This affects late-life immigrants more than immigrants who arrived during prime working ages or younger due to their greater reliance on these programs. Overall, we find that late-age immigrants incur a hefty income penalty compared with seniors who immigrated at younger ages, and that this late-age arrival penalty is greater for immigrants who arrived after the 1996 welfare reforms restricted access to SSI and Medicaid.

When combined with evidence that late-age immigrants are a growing proportion of the legal U.S. immigration stream (Carr & Tienda, 2013), our results argue for greater consideration of AAI by social science researchers and in public policy conversations about immigration reform, where it has long been absent. The consequences of a policy regime that welcomes older immigrants, but provides only limited public support after admission, warrant further examination both of admission criteria and the wisdom of curtailing benefits for admitted immigrants, irrespective of age. One policy option is to revisit, at the state and federal level, the sharp division created between citizens and LPRs created by PRWORA and IIRIRA.

The postrecession budgetary climate coupled with population aging may also lead to reconsideration of the inclusion of parents of U.S. citizens among immediate family members who are not subject to numerical limitation. Limiting the numbers of visas issued in the parent and adult sibling family reunification streams and/or introducing age preference in employment visas are two strategies to curtail late-age immigration. However, there are also more creative ways to balance budgetary priorities with the humanitarian desire to reunify families. These include imposing a “balance of kin” test that would allow sponsorship of parents only if the majority of their offspring are settled in the United States; raising the income threshold for sponsoring relatives; rigorously enforcing the affidavit of support as stipulated by IIRIRA; expanding integration of public pension systems globally through additional totalization agreements; and requiring sponsors to purchase health insurance for their elderly relatives. Each of these policy strategies requires a cost-benefit evaluation prior to enactment, as well as robust discussion of what they imply for the role of immigration in American life.

Funding

This work was supported by the Princeton Center for the Demography of Aging (National Institutes of Health Grant P30 AG024361). Additional institutional support was provided by a grant (R24HD047879) from the Eunice Kennedy Shriver National Institute of Child Health and Human Development to the Office of Population Research.

Acknowledgments

This paper was presented at the 2012 Annual Meetings of the Population Association of America, San Francisco, May 1–3. K. O’Neil and M. Tienda jointly planned the analysis and wrote and revised the paper. K. O’Neil conducted data analysis. M. Tienda conducted literature review and background research.

References

Author notes