How Are Immigrants Faring After Welfare Reform?. Conclusions


The greatest need for food stamps, as measured by food insecurity, is among LEP families and those with more children. Single-parent families also have higher poverty and food insecurity rates. All of these types of families are significantly more likely to receive food stamps, suggesting benefits are properly targeted. Yet the low rate of food stamp participation overall among food insecure immigrant families suggests there is substantial unmet need in both cities.

Unmet need appears to be highest among food insecure and low-income families not participating in TANF, SSI, general assistance or any other public benefit programs. Food stamp receipt rates are very high for TANF, suggesting that the link to welfare benefits provides much greater access to food stamps. These findings also suggest outreach to low-income immigrant communities in both cities could yield higher FSP participation among working families not participating in welfare programs.

Refugees have relatively high levels of food stamp participation in New York City. Access to benefits there is enhanced by the strong presence of private refugee resettlement and service agencies.

LEP families are more likely to participate in FSP than their English proficient counterparts, even when controlling for poverty and food insecurity. Responses to the in-depth survey provide anecdotal evidence that miscommunication between caseworkers and recipients can lead to benefit denials and delays in recertification. Nonetheless, responses to the larger survey suggest that a greater share of needy LEP compared to proficient immigrants are able to gain access to benefits. It may be that food stamp offices in both Los Angeles and New York have sufficient bilingual staff and interpretation resources to assist clients during application and recertification processes. LANYCIS does not, however, directly address language access issues. Further research into the delivery of health and human services in both cities would be necessary to address language access more fully.(51)

The reasons why most families in the survey lost benefits since 1996 include jobs, income and family composition. Thus in our sample, declines in food stamp participation have more to do with improvements in immigrants' economic well-being than with welfare reform or other policy changes. It is important to keep in mind, however, that our sample was taken in Los Angeles and New York, two large cities with relatively generous benefits and large numbers of immigrants when compared to other localities. Therefore, when compared to immigrants living in other jurisdictions, immigrants in these cities may have been less likely to lose eligibility for benefits due to policy changes.

LANYCIS was conducted in two cities with relatively generous eligibility rules for noncitizens, due to supplemental food assistance programs funded by the states of California and New York. Both cities also have relatively strong social service delivery infrastructures, including organizations targeting services to refugees and other immigrant populations. Thus our findings here may be conservative in comparison to conditions confronting immigrant families elsewhere in the country.


30.  The Medicare premiums cited are for 2001.

31.  Food Stamps Act of 1977, as amended. SEC. 2. (7 U.S.C. 2011). Available at

32.  Personal Responsibility and Work Opportunity Reconciliation Act, P.L. 104-193 (1996).

33.  Other groups exempted from the bar on eligibility include asylees, Amerasians and Cuban/Haitians (for five years), as well as active-duty military, veterans, and their dependents.

34.  Agriculture, Research, Extension, and Education Reform Act, P.L. 105-185 (1998).

35.  S4863, passed by the New York State Assembly and signed into law during July 2001, made this program permanent. A copy of the bill is available at

36.  An additional issue is that in California, food stamp benefits for SSI recipients are "cashed out," so that the value of food stamp benefits is added to the SSI check, rather than being issued as a separate food benefit. Many aged and disabled people, including immigrants, are getting food stamp benefits, but may be unaware of it since it simply increases the value of their SSI check.

37.  We use the term "moderate hunger" instead of hunger because the 6-question short scale we use only includes questions addressing conditions associated with moderate hunger, not the conditions of more severe hunger addressed in the longer 18-item scale (Appendix 3).

38.  The initial set of weights accounted for the stratified sampling design. The goal of the sampling design was to attain roughly equal sample sizes (number of responding households) in both cities, such that about half the unweighted sample households received food stamps in 1996 or 1997, about one-quarter had income below 200 percent of poverty but did not get food stamps in 1996-97, and one-quarter had incomes over 200 percent of poverty.

39.  Post-stratification weights brought the number of families up to totals in a pooled sample of the March 1997-99 CPS. LANYCIS family totals were adjusted to match the pooled CPS for four factors: (1) family composition, (2) poverty level, (3) country of origin and (4) schooling of the family respondent.

40.  An undocumented family includes at least one undocumented adult. In a legal immigrant family there is at least one legal immigrant adult but no undocumented adults. Refugee families are those with at least one refugee but no undocumented or legal immigrant adults. Finally, naturalized families include only naturalized adults. The legal status and citizenship of children are not considered in this classification.

41.  English, Spanish, Russian, Vietnamese and Mandarin Chinese.

42.  Respondents were first asked if they primarily speak a language other than English at home. Those who primarily speak another language (the vast majority of samples in both cities) were then asked whether they speak English "very well", "well", "not well" or "not at all." We categorize people speaking only English or English very well as proficient, and those speaking English well, not well or not at all as limited English proficient (LEP).

43.  Families with members who are only either native-born citizens or temporary nonimmigrants such as tourists, students or temporary workers (of which there are a very small number in LANYCIS) are excluded from the analyses in Part II of this report.

44.  LANYCIS is larger and focuses more closely on food stamp recipients and former recipients. Excluding families with only temporary nonimmigrant members, LANYCIS has a sample size of 3,363 immigrant families, compared to 1,707 immigrant families in the April 1999 CPS. LANYCIS also has a higher share of low-income families within its sample. The April CPS is not designed to be truly representative of the population below the state level. LANYCIS was conducted using both a food stamp recipient list and random-digit dialing samples and thus may have been more likely to reach food insecure families in the food stamp list sample. LANYCIS population totals, however, were adjusted to match March 1999 CPS population totals using weights that account for poverty, educational attainment, country of origin, and family composition.

45.  The odds-ratios and probability values displayed in Table 2.9 were generated in STATA using a logistic procedure. Dummy variables for citizenship and legal status were included with and without interactions with city (New York versus Los Angeles). Tenure in the U.S. was introduced into the model as a continuous variable and as dummies for 5, 7, 10 and 20 years. None of these variations made status or tenure significant in these models.

46.  Appendix 4 provides details concerning calculation of food stamps receipt during the previous year in the LANYCIS data.

47.  Sample sizes for food insecure refugee families without elderly members are too small to calculate food stamp receipt rates.

48.  These models were also generated by STATA using logistic regression. The coefficients and odds ratios for refugees in New York are not statistically significant, due to the small number of refugees in the sample. Yet the very high odds ratios suggest substantive significance.

49.  Nonresponse to this question was high at 14 percent in Los Angeles and 15 percent in New York.

50.  In fact, since data are for total numbers of families participating in FSP at two points in time, they may show a decrease in new applications as well as a large number of terminations.

51.  The Urban Institute is currently conducting a study of whether language or other barriers may impede immigrant access to food stamps and other federal means-tested benefits, as well as best practices to overcome any barriers. The U.S. Department of Health and Human Services, Office of the Assistant Secretary for Planning and Evaluation, is providing support for this project.


View full report


"report.pdf" (pdf, 2.23Mb)

Note: Documents in PDF format require the Adobe Acrobat Reader®. If you experience problems with PDF documents, please download the latest version of the Reader®