Welfare Reform and Older Immigrants: Food Stamp Program Participation and Food Insecurity

The Gerontologist Vol. 48, No. 1, 42 50 Copyright 2008 by The Gerontological Society of America Welfare Reform and Older Immigrants: Food Stamp Program Participation and Food Insecurity Purpose: The welfare reform bill of 1996 severely constrained noncitizens eligibility for the Food Stamp Program (FSP). This study examined the effects of eligibility restrictions on older immigrants FSP participation and food insecurity. We paid special attention to household composition and household eligibility as well as older immigrants individual eligibility. Design and Methods: The sample consisted of 3,175 low-income older adults from the 1999 Current Population Survey. We used probit and tobit regressions. We compared three analytical models: two models using the differences-in-differences approach with two different measures of immigration status (individual and household status), and one using a summary variable of the percentage of FSPeligible people in a household. Results: Both household immigration status and older adults individual immigration status were significantly associated with the probability of FSP participation. Living in a household with a higher percentage of eligible members significantly increased older people s FSP participation and benefit levels while significantly reducing their risk of food insecurity. Implications: Noncitizen eligibility restriction affects older immigrants FSP participation and food insecurity by its relation to household composition. Experts should therefore consider household composition when developing social policies for older immigrants. This study was supported in part by grants from the John A. Hartford Foundation Geriatric Social Work Faculty Scholars Program, the Center for Aging at Washington University in St. Louis, and the Faculty Research Fund from the George Warren Brown School of Social Work at Washington University in St. Louis. We are grateful to Drs. Namkee Choi and Nancy Morrow-Howell for thoughtful comments on earlier drafts of this article and to Charles Jarrett, Susanna Johnson, and Carissa Clark for their wonderful research assistance. Address correspondence to Yunju Nam, PhD, George Warren Brown School of Social Work, Washington University in St. Louis, One Brookings Drive, #1196, Saint Louis, MO 63130. E-mail: ynam@ wustl.edu 1 George Warren Brown School of Social Work, Washington University in St. Louis, MO. Yunju Nam, PhD, 1 and Hyo Jin Jung, MSW 1 Key Words: PRWORA, Eligibility restrictions, Household composition, Citizenship status, Federal and state policy A substantial proportion of older immigrants, especially those without U.S. citizenship, live in poverty. The poverty rate was approximately 20% among older noncitizens for the years 1995 to 2002 (U.S. Census Bureau, 1995, 1996, 1997, 1998, 1999, 2002). Comparable statistics for older adults as a whole were approximately 10% for the same period (He, Sengupta, Velkoff, & DeBarros, 2005). As a reflection of their poor economic status, during this period older noncitizens relied on government assistance more heavily than did their citizen counterparts. In 1994, for example, 21% of older noncitizens participated in cash assistance programs, whereas only 4% of citizens did so (Fix & Passel, 1999; Hu, 1998). Considering this high welfare program participation rate, strict eligibility restrictions toward noncitizens imposed by the Personal Responsibility and Work Opportunity Reconciliation Act (PRWORA) may have had adverse effects on older immigrants. When passed in 1996, PRWORA terminated noncitizens eligibility for the federal Food Stamp Program (FSP) and Supplemental Security Income, prohibited states from providing federally funded Medicaid and Temporary Assistance to Needy Families to noncitizens who immigrated after welfare reform (postenactment immigrants) for their first 5 years in the United States, and granted states the right to determine Medicaid and Temporary Assistance to Needy Families eligibility for immigrants who had immigrated before welfare reform (preenactment immigrants). Eligibility after PRWORA is thus determined by citizenship status and date of entry into the United States (U.S. Department of Agriculture, 2003; Zimmermann & Tumlin, 1999). Recognizing the harsh conditions imposed by PRWORA, some states provided state-funded FSPs and other public benefits to ineligible noncitizens. 42 The Gerontologist

Congress also relaxed federal eligibility rules through subsequent legislations. First, it restored federal FSP eligibility for older (aged 65 years or older at the time of welfare reform), minor (younger than age 18), and disabled immigrants in 1998. Then it reestablished working-age noncitizens eligibility in 2002 if they had lived in the United States for more than 5 years. However, citizenship status, date of entry, and state of residence remain issues for noncitizens regarding FSP benefits because most postenactment immigrants remained ineligible under the federal rules during the 5-year ban (U.S. Department of Agriculture, 2003; Zimmermann & Tumlin, 1999). Existing studies have shown that public assistance program participation rates have decreased more rapidly among immigrants, even those who have remained eligible, after welfare reform than have those of U.S. natives (Borjas, 2004; Fix & Passel, 1999). However, how much of the decline is attributable to the chilling effects of PRWORA (e.g., immigrants fear of being accused of being a public charge; Fix & Passel, 1999), improved macroeconomic conditions (Haider, Schoeni, Bao, & Danielson, 2004), or increased naturalization rates (Van Hook, 2003) is unclear. Despite a sharp decline in welfare program participation, only a limited number of studies have examined the effects of welfare reform on immigrants lack of well-being, specifically food insecurity. In fact, only two studies have focused on food insecurity among immigrants (Borjas, 2004; Van Hook & Balistreri, 2006). Borjas (2004) showed that generous state welfare eligibility rules significantly reduced noncitizens risk of food insecurity. Van Hook and Balistreri (2006) reported a higher level of food insecurity among children of noncitizens than among children of naturalized citizens, after controlling for socioeconomic characteristics. Invaluable as they are, studies in the literature have limitations. First, except for a small number of descriptive analyses (Fix & Passel, 1999; U.S. Department of Agriculture, 1999), no study has conducted separate analyses on older immigrants. Findings from previous studies therefore may not reflect the unique position of older immigrants in the welfare system and labor market. On one hand, older immigrants may have had a more difficult time responding to eligibility changes than their younger counterparts (e.g., because of increased challenges in learning English, unfamiliarity with American culture; Treas & Mazumdar, 2002). On the other hand, at times, the welfare system has treated older immigrants more favorably than their working-age counterparts. For instance, Congress restored pre-enactment older noncitizens federal FSP eligibility sooner than it did their younger counterparts. Some states have provided welfare benefits and naturalization process assistance only for older noncitizens (U.S. Department of Agriculture, 2003; Zimmermann & Tumlin, 1999). Second, previous studies have paid little attention to household composition in studying the effects of Vol. 48, No. 1, 2008 43 welfare reform on immigrants. Most studies relied solely on the immigration status of heads of households (Borjas, 2004; Haider et al., 2004), even though many households are composed of individuals with different citizenship statuses (Fix & Zimmermann, 2001). Furthermore, a household is a critical institution for survival among immigrants; resources are collected and consumed communally, and adapting to structural changes created by the host society is eased (Kibria, 1993). Reflecting this reality, older immigrants are more likely than their native counterparts to live in extended families and in larger households (Glick & Van Hook, 2002). In addition, living arrangement is correlated with older immigrants eligibility and participation in welfare programs (Van Hook, 2000). Accordingly, household composition is a critical factor in analyzing welfare program participation among older immigrants. To fill in gaps in previous studies and facilitate future policy debates, we examined FSP participation and food insecurity among older immigrants after welfare reform, paying special attention to the interaction between eligibility rules and household composition. Due to the fact that older preenactment immigrants became eligible for FSP in 1998 under the federal rules, they have equal access to FSP regardless of their state of residence. Thus, we did not expect that state generosity, or a lack thereof, toward working-age noncitizens would affect an older immigrant if the individual s eligibility was the only marker of food-related outcomes. If that were not the case, and if eligibility of other household members also played a role, state eligibility rules would likely affect older immigrants FSP participation and food insecurity. To investigate the relations among immigration status, household composition, and FSP-related outcomes among older noncitizens after welfare reform, we attempted to answer the following questions with three different measures of immigration status: Do state FSP eligibility rules affect FSP participation and food insecurity of older people on the basis of (a) their individual immigration status or (b) their household immigration status? and (c) Does the percentage of FSP-eligible people in a household explain older people s FSP participation and food insecurity? We focused on FSP because, as previously mentioned, the change in federal eligibility rules treats noncitizens of different ages unequally. Accordingly, FSP is an ideal program to study to determine whether household composition is one of the mediating factors through which eligibility rules influence the target population. Methods Data Sources and Sample We used individual data from the 1999 Current Population Survey (CPS) Annual Social and

Economic (ASEC) Supplement and the Food Security (FS) Supplement. The CPS collects labor force participation information from a nationally representative sample of the civilian, noninstitutionalized population every month. The ASEC supplemental survey is conducted annually in March to obtain detailed information on public assistance program participation. The ASEC supplement provides information on demographics, immigration status, household FSP participation, and other economic indicators. The FS supplement data, collected in April of 1999, provided household information on food insecurity. Each supplement used in this study provided information for the 12-month period preceding the interview (U.S. Census Bureau, 2001, 2006). Therefore, our data covered the period between March 1998 and February 1999 for FSP participation and the period between April 1998 and March 1999 for food insecurity. The CPS has advantages over other national data sets. It has a better measure of immigration status than the Health and Retirement Study, which does not include information on citizenship status. Furthermore, the CPS has a much larger sample than the Survey of Income and Program Participation. We chose the 1999 CPS data over the data reported in other years because of the time period covered (March 1998 to April 1999), when federal FSP eligibility rules began to discriminate workingage noncitizens from older noncitizens. We used state-level data based on two sources: (a) Zimmermann and Tumlin (1999) for policy variables related to state-funded food assistance programs for noncitizens and (b) the Bureau of Labor Statistics for state unemployment rates. We merged these state-level data into individual data, extracted from the CPS. The sample used for this study consisted of older adults (aged 65 or older) included in both the ASEC and the FS supplements. Following previous studies on food insecurity (Gundersen & Oliveira, 2001), we included only those living in low-income households (income-to-needs ratio less than 2) and those without missing values for the food security variable in the sample. We excluded eight cases of postenactment immigrants from the sample because they remained ineligible for FSP under the federal rules. We also excluded 32 American Indians because their position in the welfare system was different from that of other populations, especially if they lived on reservations (Stromwall, Brzuzy, Sharp, & Andersen, 1998). The final sample consisted of 3,175 older adults (2,886 natives, 186 naturalized citizens, and 103 noncitizens). To check the robustness of our findings, we ran separate analyses with a sample after excluding 13 cases of likely refugees (those from refugee-sending countries as defined in Passel & Clark, 1998). Analyses based on this sample produced results similar to those reported in this study. Variables and Measures We used three dependent variables: (a) FSP participation, (b) per capita food stamp benefit amount, and (c) food insecurity indicator. We created the last variable, food insecurity indicator, based on the Food Security Scale Summary Status developed by the U.S. Department of Agriculture (Bickel, Nord, Price, Hamilton, & Cook, 2000), whose validity was confirmed for both the general population (Ohls, Radbill, & Schirm, 2001) and the older population (Nord, 2003). (See Table 1 for a more detailed description of the variables.) We developed two measures of FSP eligibility rules that were applicable only to noncitizens. The first variable measured state generosity toward noncitizens. We categorized 15 states as generous (California, Connecticut, Illinois, Maine, Massachusetts, Maryland, Minnesota, Missouri, Nebraska, New Jersey, New York, Rhode Island, Texas, Washington, and Wisconsin). The second variable was the percentage of FSP-eligible members in a household, which we constructed by using both federal and state eligibility rules and household composition. We took into account the eligibility rules that applied only to noncitizens and did not consider those that applied to both citizens and noncitizens; we considered an individual eligible if he or she was a citizen, an older or minor noncitizen (eligible under the federal rule), or a working-age noncitizen in a generous state (eligible under the state rule), even if he or she had income and assets greater than those allowed in the eligibility criteria. Analysis We ran three types of analyses. The first two sets of analyses used the first measure of FSP eligibility policy (state generosity toward noncitizens), and the third set of analyses used the second measure (the percentage of eligible household members). The first two sets of analyses took a differences-in-differences approach, as have many existing studies that have evaluated policy effects (Borjas, 2004). Differencesin-differences analyses compare outcome measures between the target and nontarget populations. We took this approach because unobserved factors may correlate with both states generosity toward noncitizens and outcome measures. For example, if states with generous FSP eligibility rules happen to have simpler FSP application and recertification procedures in general, analyses comparing the target population s outcomes between generous and nongenerous states would overestimate the effect of states eligibility rules. A differences-in-differences approach would produce more accurate estimations in the presence of these types of unobserved variables. 44 The Gerontologist

Table 1. Description of Study Variables Variable Description Dependent variables FSP participation (1) An older adult lived in a household that ever participated in FSP during the preceding 12 months; Per capita food stamp The total amount of FSP benefits for the preceding 12 months divided by the household size benefit amount Food insecurity indicator Independent variables Individual immigration status Household immigration status Generous state Percentage of household members eligible Age Age squared Female Race/ethnicity Education No. of years in the U.S. Entered U.S. at age 55 or older Married Household size No. of children No. of older people Independent living Homeownership Asset income Unemployment rate Note: FSP = Food Stamp Program. (1) An older adult lived in a household with concerns over food supply and management and/or with reduced food intake of household members for the preceding 12 months; (1) Native citizen: individual born in the U.S., in the U.S. outlying territories, or abroad to a U.S. citizen (reference group); (2) Naturalized citizen: foreign-born individual who became a citizen through naturalization; (3) Noncitizen: foreign-born individual without citizenship (1) All-citizen household: a household including only citizens (reference group); (2) Mixed household: a household including both citizens and noncitizens; (3) All-noncitizen household: a household including only noncitizens (1) The state of residence provides state-funded food stamps to noncitizens ineligible for federal FSP; Number of FSP-eligible household members under either federal or state rules divided by total number of household members (and multiplied by 100) Age of an individual Square of age of an individual (1) Female; (1) White (reference): individual reported as a non-hispanic White; (2) Black: individual reported as a non-hispanic Black; (3) Hispanic: individual reported as a Hispanic; (4) Asian: individual reported as an Asian or Pacific Islander (1) Less than high school (reference group): individual without a high school diploma; (2) High school graduate: individual with high school diploma but without further schooling; (3) College education: individual with college education Number of years since entry into the U.S.; a continuous variable is used in Table 1; In multivariate analyses, dummy variables were used: (1) Native citizen or living in the U.S. longer than 30 years (reference group); (2) Living in the U.S. 21 30 years; (3) Living in the U.S. 11 20 years; (4) Living in the U.S. for 10 years or less (1) Entered the U.S. at age 55 or older; (1) Currently married; Number of people in a household Number of children younger than 18 in a household Number of people aged 65 or older in a household (1) Living alone or only with spouse; (0) Living with someone other than spouse (1) Living in a house owned by a household member; (1) An individual living in a household who has income from assets (e.g., interest, dividends, or rents); State unemployment rate from the U.S. Bureau of Labor Statistics Vol. 48, No. 1, 2008 45

A model specification of the differences-in-differences analyses was as follows: Y i ¼ a þ I i 3 b 1 þ G i 3 b 2 þði i 3 G i Þ 3 b 3 þ X i 3 b 4 þ e i ð1þ; where Y i is the dependent variable, I i is a vector of immigration status dummies for individual i, G i indicates whether an individual i lives in a generous state, X i indicates a vector of control variables, and e i is random error. The parameters of interest were the elements in the vector b 3, coefficients of the interaction term between immigration status and state FSP eligibility rules. These coefficients estimated whether and how the FSP participation and food insecurity of older adults with different immigration statuses reacted differently to state eligibility rules. To test whether an older individual s eligibility was the only marker of food-related outcomes or whether other household members eligibility also played a role, we ran two sets of analyses with two different measures of immigration status. We based the first measure on an individual s immigration status. We categorized older adults into three groups: naturalized citizen, noncitizen, and native citizen (reference group). We created the second measure at the household level while taking into account every household member s immigration status. This measure had three categories: all-noncitizen household, consisting solely of noncitizens; mixed household, consisting of both citizens and noncitizens; and all-citizen household (reference group). Because older pre-enactment immigrants became eligible for FSP under the federal rules regardless of their citizenship status in 1998, we expected the coefficients of the interaction term between individual immigration status and state generosity (b 3 )tobe insignificant if only older individuals eligibility mattered. The coefficients of the same variable (b 3 ) should, however, have been statistically significant if working-age noncitizens eligibility affected older noncitizens (or naturalized citizens ) FSP participation and food insecurity. Using the second measure of immigration status at the household level was another way to examine whether older noncitizens FSP participation and food insecurity were affected by household composition. If the state s eligibility rules distinctly influenced older people living in different types of households, the coefficients of the interaction term between household immigration status and state generosity (b 3 ) should have been statistically significant. The last set of analyses used the second FSP eligibility measure, the percentage of household members eligible for FSP. As described, this measure considered both household composition and state and federal eligibility rules. Model specification of this type of analysis was as follows: Y i ¼ a þ E i 3 b 1 þ I i 3 b 2 þ X i 3 b 3 þ e i ð2þ; where Y i is the dependent variable, E i is the percentage of household members eligible for FSP, I i is a vector of immigration status dummies for individual i, X i is a vector of control variables, and e i is random error. In this type of analysis, we expected the coefficient of the percentage of eligible household members (b 1 ) to be significantly positive for food stamp receipt and food stamp amount and to have a significantly negative coefficient in the food insecurity analysis if household eligibility affected older adults FSP participation. We controlled for individual immigration status in case unobservable differences among native citizens, naturalized citizens, and noncitizens existed. The model without an individual immigration status dummy variable produced the same results as those reported in this study. We used probit regressions for the two dichotomous dependent variables (food stamp receipt and food insecurity) and tobit regressions in analyzing food stamp amount because a substantial proportion of the sample reported a zero value for this variable. Considering that some sample households had multiple observations when they included more than one older person, we estimated our models with robust standard errors (Greene, 2003). As shown in Equations 1 and 2, each model incorporated demographic and household characteristics, immigration-related variables, and state unemployment rates as control variables. Table 1 briefly describes these control variables. To check the robustness of the analysis model, we ran several models with additional control variables: (a) a model including a household income variable in addition to the control variables in the original model, (b) a model with additional state-level variables (the percentage of foreign-born people in the state of residence in 1990, changes in the percentage of foreign-born people in the state between 1990 and 2000, and state supports to noncitizens naturalization process), (c) a model including the interaction between state unemployment rate and immigration status, and (d) a model with state dummy variables. We also ran analyses on a sample that excluded California residents, in order to check whether special conditions in California explained rapid declines in welfare program participation among noncitizens, as Borjas (2001) argued. The results from these analyses did not differ substantially from those reported in this study. As the U.S. Census Bureau (2001, 2006) instructed, we weighted our data with the adjusted individual weight variable included in FS supplemental data. The Census Bureau developed this weight variable to obtain representative estimates when analyzing the sample limited to those with valid values for Food Security Scale variables. 46 The Gerontologist

Table 2. Characteristics of Sample by Immigration Status (Weighted) Variable Native Naturalized Noncitizen Total Age (years) a,b 75.85 75.41 73.40 75.73 Female (%) 67.48 65.14 67.49 67.33 Race/ethnicity (%) a,b,c Non-Hispanic White 81.67 51.39 28.67 77.71 Non-Hispanic Black 14.05 4.74 4.15 13.07 Hispanic 3.95 31.30 46.02 7.31 Asian/Pacific Islander 0.33 12.57 21.15 1.91 Education (%) a Less than high school 47.88 49.84 63.54 48.60 High school graduate 32.35 28.18 19.68 31.61 College 19.77 21.98 16.79 19.80 Owning a house (%) a,b,c 73.71 58.66 40.12 71.47 Having asset income (%) a,b 47.51 44.46 31.77 46.71 Average household size a,b 1.70 1.91 2.58 1.75 Average no. of older people in household 1.36 1.37 1.30 1.36 Average no. of children in household a,b 0.08 0.16 0.54 0.11 Married (%) 37.24 44.98 45.42 38.04 Living independently (%) a,b 80.24 81.48 59.88 79.55 Entered U.S. after age 55 (%) b 14.08 48.64 26.89 Average no. of years in the U.S. b 37.68 21.64 31.74 Living in a mixed household (%) a,b,c 0.43 10.25 40.21 2.57 Average percentage of household members eligible a,b,c 99.96 98.58 88.87 99.45 Living in a generous state (%) a,c 40.49 62.85 69.73 43.04 Average state unemployment rate a,c 4.48 4.92 5.05 4.53 Unweighted n (weighted %) 2,886 (89.77%) 186 (6.44%) 103 (3.79%) 3,175 (100.00%) Notes: a Difference between native and noncitizens significant at the.05 level. b Difference between naturalized citizens and noncitizens significant at the.05 level. Difference between native and naturalized citizens significant at the.05 level. Results Descriptive Statistics Table 2 summarizes demographic and socioeconomic characteristics of the sample by individual immigration status. Older noncitizens differed significantly from citizens, especially from native citizens, in socioeconomic characteristics. On average, noncitizens had lower levels of education and were less likely to own assets (house or asset income) than native or naturalized citizens. They were less likely to live independently and were more likely to live in mixed households. Table 3 clearly shows differences in FSP participation and food insecurity by immigration status. Older noncitizens were significantly more likely to participate in FSP than native and naturalized citizens. Despite higher FSP participation rates and benefit levels, noncitizens were significantly more likely to experience food insecurity than the other groups. Multivariate Analysis Table 4 summarizes the results from multivariate analyses. Model 1 reports results based on individual immigration status, Model 2 describes findings from the model using household immigration status, and Model 3 summarizes findings based on the percentage of eligible household members variable. As previously described, each regression controlled for potentially confounding factors, such as demographic and household characteristics in the model. These control variables had expected relations with the dependent variable. Education and asset ownership (homeownership and asset income) significantly reduced the probability of receiving food stamps and food insecurity. (Full analysis results are available upon request.) FSP Participation. Table 4 shows that residence in generous states significantly increased older noncitizens FSP program participation even after the federal rules restored their eligibility. The interaction term between noncitizen and state generosity had a significantly positive coefficient at the.05 level, indicating that states generosity toward working-age noncitizens significantly increased older noncitizens FSP participation. Results from Model 2 showed that household immigration status was significantly associated with older adults likelihood of receiving food stamps, supporting the hypothesis that state generosity affects older noncitizens through its interaction with household composition. The interaction term between mixed household and state generosity was significantly positive at the.1 level, implying that states generous FSP eligibility rules may have increased FSP participation among older adults Vol. 48, No. 1, 2008 47

Variable Table 3. FSP Participation and Food Insecurity Experience of Sample Native Naturalized Noncitizens Total FSP participation (%) a,b 8.12 7.35 22.70 8.62 FSP benefit amount ($) a,b 31.99 48.12 190.93 39.01 Food insecurity (%) a,b 10.19 10.15 17.01 10.45 Notes: FSP = Food Stamp Program. a Difference between native and noncitizens significant at the.05 level. b Difference between naturalized citizens and noncitizens significant at the.05 level. living in mixed households. The interaction term between an all-noncitizen household and state generosity, however, was not statistically significant, although the direction of the relation was as expected. This may be explained by the fact that households consisting solely of noncitizens are mainly composed of older people and therefore are not constrained by the citizenship requirement. According to our data, the percentage of households with at least one working-age person was 76.48% among mixed households, whereas it was estimated as only 21.37% among households composed solely of noncitizens. Model 3 also produced results indicating the significant role of household composition in FSP participation. The percentage of eligible individuals in the household variable, which captured both FSP eligibility rules and the household composition, had a significantly positive coefficient at the.05 level. This result suggested that, after controlling for relevant factors, the higher the percentage of eligible household members, the more likely older adult households were to receive food stamps. Food Stamp Benefit Amount. The associations between eligibility restrictions on noncitizens and food stamp benefit amount were not as consistent across models as those observed in the FSP participation analyses. The results from Models 1 and 2 showed that none of the interaction terms between immigration status and state generosity had a significant association with the value of food stamps received for the preceding 12 months. The percentage of eligible household members variable in Model 3, however, showed a significantly positive relation with food stamp benefit amount at the.05 level, even after individual immigration status was controlled for. This result indicated that the higher the percentage of household members eligible for FSP, the higher the per capita FSP benefit that older adults households received. Table 4. Multivariate Analysis Results: FS Eligibility and Older Immigrants Variable FSP Participation FSP Amount Food Insecurity Model 1: Individual immigration status State generosity 0.371*** 120.908 0.247*** 0.095 87.570 0.083 Naturalized citizen 0.433 133.738 0.220 0.286 273.364 0.273 Noncitizen 0.698 140.423 0.116 0.446 372.507 0.351 Generosity 3 Naturalized citizens 0.368 144.469 0.194 Generosity 3 Noncitizens 0.354 359.687 0.313 0.963** 312.905 0.059 0.425 459.214 0.375 Model 2: Household immigration status State generosity 0.360*** 113.709 0.232*** 0.092 83.033 0.080 Mixed household 1.398* 1003.113* 0.049 0.731 568.445 0.413 All noncitizens 0.277 254.706 0.179 0.434 469.426 0.450 Generosity 3 Mixed household 1.201* 731.179 0.275 Generosity 3 All noncitizens 0.728 567.990 0.474 0.726 123.682 0.056 0.470 539.704 0.495 Model 3: Percentage of eligible people in a household Percentage eligible 0.024** 19.296** 0.016** 0.010 9.785 0.007 Naturalized citizens 0.241 57.491 0.127 0.198 215.799 0.173 Noncitizen 0.080 191.448 0.336 0.268 214.367 0.265 Notes: In addition to variables reported in this table, this study controlled for demographic and household characteristics in each analysis. FSP = Food Stamp Program. *p,.1; **p,.05; ***p,.01. Food Insecurity. In the food insecurity analyses, state generosity did not make a significant difference in food insecurity among older adults with different immigration statuses measured at either the individual or the household level. None of the interaction terms between state generosity and immigration status had a significant coefficient in either Model 1 or Model 2. Results from Model 3, however, illustrated that a high percentage of eligible household members significantly reduced an older adult s risk of food insecurity. Discussion The PRWORA of 1996 imposed a barrier to immigrants welfare program participation by restricting noncitizens eligibility. In contrast with the pre-prwora era, when every lawful permanent resident had the same rights in the welfare system as citizens, the post-prwora era is marked by discriminatory treatment of immigrants based on their citizenship status, age, date of entry into the United States, and state of residence. We examined the effects of these changes in welfare rules on older immigrants FSP participation 48 The Gerontologist

and food insecurity, adding important new insight into understanding the effects of welfare reform on older immigrants. Specifically, we considered the role of household composition in FSP participation and food insecurity. Our findings suggest that relaxation of the federal FSP eligibility rules of 1998 may not have benefited older noncitizens equally because older adults FSP participation is affected by the eligibility of other household members. As such, we expanded researchers understanding of immigrants welfare program participation and their well-being in the period since welfare reform by adding household composition to the list of mediating factors in addition to the chilling effects (Fix & Passel, 1999), shifts in naturalization rates (Van Hook, 2003), and macroeconomic conditions (Haider et al., 2004) discussed in existing literature. What is not clear, however, is why eligibility of other household members affects FSP participation by eligible older immigrants despite the fact that eligible individuals can apply for FSP and receive benefits. We may explain this by the low level of benefits available when only a small number of household members is eligible. Existing studies on FSP participation in the general population have shown that eligible households often decide not to apply for FSP when the expected benefit levels are low, because they view it as not being worth the time and effort (McConnell & Ponza, 1999). It is understandable, then, that an immigrant household would decide not to apply for FSP when only older and minor members are eligible, and therefore the expected benefit level seems too low to bother applying. In addition, households with ineligible members, especially those including working-age noncitizens, may be reluctant to apply for public benefits for fear of complicating their future citizenship applications. This explanation is, however, speculative and requires further investigation. Older noncitizens experience food insecurity at a significantly higher rate than older native and naturalized citizens, despite their higher FSP participation rates and higher benefit levels. This finding illustrates the dire economic conditions of older noncitizens; they have a difficult time meeting their basic dietary needs even when FSP is available. This situation reflects their economic and social disadvantages (e.g., low level of education and asset ownership). This study is not free of limitations. First, we utilized a cross-sectional data set and were therefore unable to trace changes in FSP participation and food insecurity of individual immigrants over time. In particular, we were unable to test whether mediating effects of household composition were attributable to differences in living arrangements that had existed before welfare reform (Glick & Van Hook, 2002) or to immigrant households rearrangement of their composition as a response to eligibility restrictions. Vol. 48, No. 1, 2008 49 Second, data limitations did not allow us to completely resolve the issue of unobserved covariates. For example, it was not possible to determine whether different state eligibility rules reflected the attitudes of state food stamp offices toward noncitizens or the work of active immigrant advocacy groups. Further research is needed to sort out these potentially confounding factors from eligibility rules. Third, we cannot rule out the possibility that the CPS s own limitation in collecting information from non-english-speaking populations may have affected our results. The CPS data were collected without any specific and systematic guidelines for field interviewers when using interpreters to interview non-english-speaking households (Pan, 2005). Considering that immigrants are more likely to be non-english speakers than are natives, CPS s inattention to interview language issues may have affected analysis results. Finally, we were unable to study postenactment immigrants who still remained ineligible for FSP and other public program benefits in most states during the 5-year ban. The small number of postenactment immigrants in the data (n = 8), however, made reliable analysis impossible to conduct. The lack of empirical studies calls for future studies on this vulnerable population. This study has several implications for future policy development and research. The findings urge the consideration of household composition and living arrangements of older immigrants when developing social policies. The study also confirms that the household plays an important economic role in the lives of immigrants. Policies based on a nuclear family model therefore may not work effectively with older immigrants. At the same time, experts should reconsider current restrictions on the welfare program participation of postenactment immigrants. Immigrants often help extended kin in need and incorporate newly immigrated family members into their homes (Kibria, 1993), even if such action results in economic hardship. Our results also have implications for future gerontology research. First, they show the importance of contextual factors, such as household composition and living arrangement, in studying older immigrants. Future research should recognize the importance of contextual factors and conduct household-level analyses as well as individual-level analyses, when needed. Second, developing policy measures that reflect various aspects of a given policy is important. Different measures often produce distinct results and help researchers understand issues from various perspectives. State generosity toward noncitizens succeeds in capturing only the effects on older noncitizens FSP participation, whereas the percentage of eligible household members variable illustrates policy effects on all three outcome measures.

This result suggests that using various policy measures may help capture diverse aspects of a policy and expand knowledge of policy effects by broadening perspectives. References Bickel, G., Nord, M., Price, C., Hamilton, W., & Cook, J. (2000). Guide to measuring household food security. Washington, DC: U.S. Department of Agriculture. Borjas, G. J. (2001). Welfare reform and immigration. In R. M. Blank & R. Haskins (Eds.), The new world of welfare (pp. 369 385). Washington, DC: Brookings Institution Press. Borjas, G. J. (2004). Food insecurity and public assistance. Journal of Public Economics, 88, 1421 1443. Fix, M. E., & Passel, J. S. (1999). Trends in noncitizens and citizens use of public benefits following welfare reform. Washington, DC: Urban Institute. Fix, M., & Zimmermann, W. (2001). All under one roof: Mixed-status families in an era of reform. International Migration Review, 35, 397 419. Glick, J. E., & Van Hook, J. (2002). Parents coresidence with adult children: Can immigration explain racial and ethnic variation? Journal of Marriage and Family, 64, 240 253. Greene, W. H. (2003). Econometric analysis (5th ed.). Upper Saddle River, NJ: Prentice Hall. Gundersen, C., & Oliveira, V. (2001). The Food Stamp Program and food insufficiency. American Journal of Agricultural Economics, 83, 875 887. Haider, S. J., Schoeni, R. F., Bao, Y., & Danielson, C. (2004). Immigrants, welfare reform, and the economy. Journal of Policy Analysis and Management, 23, 745 764. He, W., Sengupta, M., Velkoff, V. A., & DeBarros, K. A. (2005). 65þ in the United States: 2005. Washington, DC: U.S. Census Bureau. Hu, W.-Y. (1998). Elderly immigrants on welfare. Journal of Human Resources, 33, 711 741. Kibria, N. (1993). Family tightrope: The changing lives of Vietnamese Americans. Princeton, NJ: Princeton University Press. McConnell, S., & Ponza, M. (1999). The Reaching the Working Poor and Poor Elderly study: What we learned and recommendations for future research. Washington, DC: Mathematica Policy Research. Nord, M. (2003). Measuring the food security of elderly persons. Family Economics and Nutrition Review, 15(1), 33 46. Ohls, J., Radbill, L., & Schirm, A. (2001). Household food security in the United States, 1995 1997: Technical issues and statistical report. Final report of the project to analyze 1996 and 1997 food security data. Princeton, NJ: Mathematica Policy Research. Pan, Y. (2005). Development of guidelines on the use of interpreters in survey interviews. Washington, DC: U.S. Census Bureau. Passel, J. S., & Clark, R. L. (1998). Immigrants in New York: Their legal status, income, and taxes. Washington, DC: Urban Institute. Stromwall, L. K., Brzuzy, S., Sharp, P., & Andersen, C. (1998). The implications of welfare reform for American Indian families and communities. Journal of Poverty, 2(4), 1 16. Treas, J., & Mazumdar, S. (2002). Older people in America s immigrant families: Dilemmas of dependence, integration, and isolation. Journal of Aging Studies, 16, 243 258. U.S. Census Bureau. (2001). Current population study, April 1999: Food security supplement. Washington, DC: Author. U.S. Census Bureau. (2006). Current population study, March 1999. Washington, DC: Author. U.S. Census Bureau. (1995, 1996, 1997, 1998, 1999, 2002). Current Population Survey (CPS) reports: Foreign-born population in the U.S., detailed tables Washington, DC: Author. U.S. Department of Agriculture. (1999). Who is leaving the Food Stamp Program? An analysis of caseload changes from 1994 to 1997. Washington, DC: Author. U.S. Department of Agriculture. (2003). Non-citizen requirements in the Food Stamp Program. Washington, DC: Author. Van Hook, J. (2000). SSI eligibility and participation among elderly naturalized citizens and noncitizens. Social Science Research, 29, 51 69. Van Hook, J. (2003). Welfare reform s chilling effects on noncitizens: Changes in noncitizen welfare recipiency or shifts in citizenship status? Social Science Quarterly, 84, 613 631. Van Hook, J., & Balistreri, K. S. (2006). Ineligible parents, eligible children: Food stamps receipt, allotments, and food insecurity among children of immigrants. Social Science Research, 35, 228 251. Zimmermann, W., & Tumlin, K. C. (1999). Patchwork policies: State assistance for immigrants under welfare reform. Washington, DC: Urban Institute. Received February 15, 2007 Accepted June 29, 2007 Decision Editor: William J. McAuley, PhD 50 The Gerontologist