PUBLIC ASSISTANCE USE AMONG YOUNG ADULTS: VARIATIONS BY PARENTAL NATIVITY Lydia R. Anderson A Thesis Submitted to the Graduate College of Bowling Green State University in partial fulfillment of the requirements for the degree of MASTER OF ARTS May 2017 Committee: Kelly S. Balistreri, Advisor Wendy D. Manning Kara Joyner
ABSTRACT ii Kelly S. Balistreri, Advisor The economic effect of immigration on the United States economy and its cost to American citizens has long been of public policy concern. Particularly, there is concern about immigrants taking advantage of the United States welfare system, which in part led to the nativity-based restrictions in the Personal Responsibility Work Opportunity and Reconciliation Act in 1996. However, much of the research on the welfare receipt of the children of immigrants is based on data collected during the 1990s or is based on select geographic samples, and therefore does not represent the newest waves of immigrants and their children. Using data from the 2012-2016 Current Population Survey, Annual Social and Economic Supplement, this study presents the current public assistance use of a nationally representative sample of young adult (25-29) children of immigrants and the children of natives. Additionally, using data from the 2001 American Community Survey, this study utilizes a mean grouping estimation strategy to examine the intergenerational relationship between parent program participation and young adult program participation and socioeconomic achievement. This study finds that among those living at or below 185% of the federal poverty line, the young adult children of immigrants have lower levels of cash assistance, food stamp, and Medicaid receipt than the young adult children of natives. In particular, the young adult children of Central American immigrants are less likely to utilize public assistance programs than the young adult children of natives. The results also suggest that there is no intergenerational transmission of public assistance use between a cohort of immigrant parents and a cohort of the young adult children of immigrants, but that immigrant
parents may use public assistance programs as a way to invest in their children s educational iii attainment.
ACKNOWLEDGMENTS iv First, I would like to thank my advisor, Dr. Kelly Balistreri for her help in writing my thesis, as well as for her encouragement and support throughout my graduate career. I consider myself very lucky to have worked with Dr. Balistreri my first semester, and am grateful for her guidance as well as her constant positive, go get em attitude. I would also like to thank my committee members, Drs. Wendy Manning and Kara Joyner for their invaluable feedback and encouragement. Additionally, I would like to thank the faculty, staff, and my fellow graduate students at the Center for Family and Demographic Research and the National Center for Family and Marriage Research. Working as a graduate assistant at the CFDR/NCFMR has been one of my favorite parts of graduate school, and I have learned so much from this experience. Finally, I would like to thank my friends and family for their continued support and encouragement. My parents, Sue and Eric Anderson, have always encouraged me to reach higher and have provided much support throughout this process. Thank you to Joe, for always encouraging me and helping to keep me motivated. To all of my friends, both near and far, thank you for your continued love and support, and for making this process a little easier with your kind words and the laughter we have shared.
v TABLE OF CONTENTS Page INTRODUCTION... 1 BACKGROUND... 6 The Children of Immigrants... 6 Poverty among the Children of Immigrants... 6 Welfare Reform and its Effect on Immigrant Program Participation... 7 Public Assistance Use among the Children of Immigrants... 10 Intergenerational Transmission of Welfare Use... 12 THE CURRENT STUDY... 15 DATA AND METHODS... 18 Current Population Survey... 18 Low-Income Sample... 20 American Community Survey... 20 Dependent Variables... 21 Independent Variable... 22 Control Variables... 22 ANALYSIS PLAN... 25 RESULTS... 28 Descriptive Results... 28 Logistic Regression Predicting Low-Income Status... 29 Public Assistance Use... 30 Cash Assistance... 30
vi Food Stamps... 31 Medicaid... 32 Any Public Assistance Use... 33 Intergenerational Correlations in Welfare Use... 34 Generational Comparison... 34 Intergenerational Correlation... 35 DISCUSSION... 38 REFERENCES... 42
vii LIST OF TABLES Table Page 1 Demographic and Socioeconomic Characteristics of Young Adults Aged 25-29 by Parental Nativity Status, 2012-2016... 49 2 Demographic and Socioeconomic Characteristics of Low-Income (Under 185% of Poverty Line) Young Adults Aged 25-29 by Parental Nativity Status, 2012-2016... 51 3 Results from Logistic Regression Models Predicting Low-Income Status of Young Adults Aged 25-29 (185% or Less of Poverty Line)... 53 4 Results from Logistic Regression Models Predicting Cash Assistance Use of Low- Income (185% or Less of Poverty Line) Young Adults Aged 25-29... 54 5 Results from Logistic Regression Models Predicting Food Stamp Use of Low-Income (185% or Less of Poverty Line) Young Adults Aged 25-29... 55 6 Results from Logistic Regression Models Predicting Medicaid Use of Low-Income (185% or Less of Poverty Line) Young Adults Aged 25-29... 56 7 Results from Logistic Regression Models Predicting Any Public Assistance Use of Low- Income (185% or Less of Poverty Line) Young Adults Aged 25-29... 57 8 Demographic and Socioeconomic Characteristics of Household Heads with Children Aged 11-18 in 2001 and Young Adults Aged 25-29 in 2012-2016... 58 9 Estimated Coefficient of Transmission Between Parents Public Assistance Use and Second Generation s Public Assistance Use... 60
1 INTRODUCTION The United States has a long history as a nation of immigrants. In 2015, it was home to 47 million immigrants, or a fifth of the world s migrant population (Menozzi 2016). Currently, 26 percent of the population in the United States is an immigrant or the child of an immigrant (Zong and Batalova 2016). The socioeconomic attainments and the integration of immigrants and their children have long been of interest to scholars and policymakers alike. Indeed, the outcomes of the children of immigrants are often seen as an indicator of the success or the failure of America s immigration policy. At the same time, a recent national poll found that 35% of Americans see immigrants as a burden to the country by taking jobs, housing, and health care (Jones 2016). While one of the major public concerns about immigration is natives losing their jobs to immigrants, another is that immigrants will take advantage of the United States welfare system. Further, there is public concern that if the U.S. provides public assistance to immigrants, they may raise their children to be reliant on welfare. This fear over the possible intergenerational transmission of welfare use is not new and led in part to the nativity-based restrictions imposed by the Personal Responsibility Work Opportunity and Reconciliation Act (PRWORA, also known as Welfare Reform) in 1996, which dramatically restricted access to public assistance for immigrant families. Prior research on the young adult children of immigrants suggests that on average they achieve higher levels of education and income than that of their immigrant parents (Crul, Zhou, Lee, Schnell, and Keskiner 2012; Hofferth and Moon 2016; Tran and Valdez 2015), and with some exceptions perform as well as or better than the children of natives (Crosnoe and Turley 2011; Hofferth and Moon 2016; Tran and Valdez 2015). While valuable, this research has two shortcomings. First, much of this research is based on young adults growing up prior to the
2 implementation of Welfare Reform and therefore may not reflect the true conditions that the children of immigrants experienced as they transitioned to adulthood. That is, the extant studies on the young adult children of immigrants are based on samples that did not face immigrantbased welfare restrictions during their childhood (e.g., Children of Immigrants Longitudinal Study, The Adolescent Health Survey) or are based on select geographic samples (e.g., Immigrant Second Generation in Metropolitan New York, Immigration and Intergenerational Mobility in Metropolitan Los Angeles). This is an important gap in the literature, given the emerging evidence which suggests that the immigrant-based restrictions of PRWORA lead to poor outcomes among young children growing up in immigrant families. For example, Kalil and Crosby (2009) found that among Latino preschool children growing up in immigrant families, those that lost access to welfare benefits after PRWORA had significantly worse health than those whose families were able to retain their access to benefits. In addition, Condon, Filindra, and Wichowsky (2015) found that Latino children of immigrants living in states that offered no supplemental support were less likely to graduate from high school post-reform than their peers living in states that had a supplemental safety net. These findings raise concern that an unintended effect of Welfare Reform on immigrant families may have a negative spillover effect on children s later socioeconomic attainment. It may be that recent cohorts of the young adult children of immigrants those whose families were targets of immigrant based welfare restrictions as children are not achieving the same levels of attainments as the children of natives. By not achieving the same levels of attainments as the children of natives, the children of immigrants could be more likely to be low-income, and therefore more likely to utilize public assistance programs.
3 Recent research using data from the post-welfare Reform period finds that among lowincome households, immigrants (with some exceptions) are less likely to use a full range of public benefits (i.e., cash assistance, food assistance, SSI) compared to native households (Bitler and Hoynes 2013; Capps, Fix, and Henderson 2009). These findings suggest that Welfare Reform had the intended effect of lowering immigrant household welfare participation, taken as a whole. However, given the dampening effect the restrictions appear to have had on child health and educational attainment, it is important to understand current patterns of welfare participation among a cohort of young adult children of immigrants. Current research has largely ignored welfare use patterns among the young adult children of immigrants. This is surprising, given that one of the biggest concerns driving the passage of Welfare Reform was the issue of immigrant family dependency on welfare programs, or a negative acculturation toward welfare and away from work (Fix, Capps, and Kaushal 2009:29). Yet, little research has examined whether there is any evidence among recent groups of young adult children of immigrants of an intergenerational correlation in welfare receipt. Research conducted prior to the enactment of Welfare Reform finds a positive and significant correlation between welfare receipt of the first and second generations. However, once education levels of the second generation are held constant, the relationship disappears, suggesting that the intergenerational transmission of the effects of welfare operate through the transmission of group skill levels (Butcher and Hu 2000; Levine and Zimmerman 1996). Other research suggests that participation in public assistance programs, rather than creating a generation of welfare dependents, might serve to support the future outcomes of the children of immigrants. Balistreri (2010) found that among young adult children of immigrants, those whose parents utilized public assistance programs when they were children had better educational and labor force outcomes
4 than those who did not. However, these studies are based on data collected prior to the immigrant-based restrictions implemented under Welfare Reform. It may be that in the postreform era, the association between immigrant parental welfare receipt and young adult socioeconomic attainments has shifted. Today s young adults experienced a unique sequence of events growing up, which could influence their public assistance use patterns. They were in their pre-teens and teenage years at the time of Welfare Reform enactment. This means that for the children of immigrants, it is possible that their families lost their public assistance eligibility while they were still living in the parental home. Additionally, they were in their late teens and early 20s finishing their education, entering the labor market, and preparing to leave the parental home during the Great Recession. This sequence of events could have economically disadvantaged today s young adults, making them more likely to be low-income and eligible for public assistance receipt. Prior research has found that key features of the public safety net, namely unemployment insurance and the Supplemental Nutrition Assistance Program (SNAP, formerly known as Food Stamps), helped to protect against of the negative economic effects of the Great Recession for some families, but not for others (Bitler and Hoynes 2015). However, the effect of Welfare Reform and the Great Recession on young adult s welfare receipt is a topic that has been largely ignored in the existing literature. Given the existing gaps in the literature, the current study will make three contributions. First, the current study updates the descriptive evidence on who is low-income among young adults (ages 25 to 29) using the most current data available. Second, the current study will provide evidence of patterns of program use among the young adult children of immigrants for the age group of young adults who may have been exposed to the immigrant-based welfare
5 restrictions during childhood. Finally, the current study will examine the intergenerational correlation between the program participation of a cohort of immigrant parents and the program participation and educational attainments of a recent cohort of young adult children of immigrants.
6 BACKGROUND The Children of Immigrants In 2013, almost one-quarter of children in the United States had at least one immigrant parent, and the number is growing. In 2006, there were 15.7 million children of immigrants, and by 2013 the number grew to 17.6 million (Woods, Hanson, Saxton, and Simms 2016). While their parents are immigrants, the overwhelming majority of children of immigrants are native born. Ninety percent of children of immigrants under age 18 were U.S. citizens, and over half had at least one parent who was a U.S. citizen in 2013 (Woods, et al. 2016). Not only do the children of immigrants comprise a large proportion of the young adult population, but they are also members of the generation that will replace what is currently the largest generation, the Baby Boomers, who are mostly native born. Therefore, their emergence into adulthood and participation in society will become increasingly important in the years to come (Reisel, Lessard- Phillips, and Kasinitz 2012). In the current study, I consider the children of immigrants as those born in the United States to at least one foreign-born parent or a foreign-born individual who arrived in the United States prior to age 12, also known as the 1.5 generation. Since the 1.5 generation spent little time in their home county and they are primarily educated and socialized in the United States, it is expected that their outcomes will be very close to those of the native-born children of immigrants (Rumbaut 2004). Poverty among the Children of Immigrants Regardless of their own nativity, children of immigrants are more likely to live in poverty than their native-born peers (Woods et al. 2016) though this varies greatly by parental country of origin (Chaudry and Fortuny 2010). In 2009, 17% of children under 18 living in native born
7 households and households with only one immigrant parent lived in poverty. By comparison, close to 30% of children with two immigrant parents lived in poverty (Borjas 2011). Among children of immigrants under 18, over half 52% lived in low-income families (under 200% of the federal poverty line), compared to 39% of children of natives in 2015 (Jiang, Granja, and Koball 2017). Poverty and low-income status among children varies greatly by race and ethnicity. Black, Hispanic, and American Indian children had the highest levels of poverty and low-income status in 2015, with over 30% of children living in poverty and over 60% of children under 18 living in low-income families. By contrast, white and Asian children had the lowest levels of poverty and low-income status in 2015, with 12% living in poverty and around 30% of children under 18 living in low-income families (Jiang, Granja, and Koball 2017). This study examines whether or not these differences in poverty based on parental nativity and race and ethnicity continue into young adulthood. Welfare Reform and its Effect on Immigrant Program Participation A majority of research studying immigrant welfare use focuses on the effects of Welfare Reform on public assistance use among immigrants, and whether or not Welfare Reform had its desired effects in reducing program participation among the foreign-born. PRWORA restricted noncitizens from receiving welfare until they had lived in the United States for at least five years, or at least ten years for food stamps. Welfare Reform also allowed states to determine their own program eligibility and to provide supplemental programs to those considered ineligible, leading to great diversity in state generosity (Borjas 2002; Potochnick 2016).
8 In 2002, the Farm Security and Rural Investment Act (2002 FARM bill) worked to reduce the food stamp restrictions imposed by Welfare Reform. It reduced the length of residence time from ten years to five years, similar to the residency requirement for other welfare programs. In addition, it also restored benefits to the disabled and gave eligibility to all legal permanent resident children without a length of residency requirement (Potochnick 2016). Although the 2002 FARM bill did restore benefits to some immigrants who were excluded under Welfare Reform, many immigrants remained confused about their eligibility and did not take advantage of the benefits. Indeed, this patchwork of state rules led to much confusion about eligibility and an avoidance of accessing welfare benefits (Zimmermann and Tumlin 1999). In addition, many immigrants, particularly those from Mexico who may have been undocumented, feared that taking advantage of government benefits would make it harder for them to gain citizenship or cause them to be deported (Potochnick 2016). There is an extensive literature on the impact of Welfare Reform on immigrants likelihood of receiving public assistance (Fix and Passel 2002), experiencing food insecurity (Van Hook and Balistreri 2006), having health insurance (Kaushal and Kaestner 2005), and living in poverty (Borjas 2011). In general, most studies found that program participation decreased for both citizen and noncitizen families after Welfare Reform, but that noncitizens experienced a greater decline (Fix and Passel 2002). Between 1994 and 1999, Temporary Assistance for Needy Families (TANF) and food stamp participation experienced the greatest declines among noncitizen families, with TANF participation decreasing by 60% and food stamp participation decreasing by 48%. Cash assistance also experienced a 44% decline, dropping from 11.2% of noncitizen families utilizing cash assistance in 1994 to 6.3% in 1999. Medicaid experienced the smallest percent decline, experiencing a 15% decrease, but still remained the
9 program with the largest participation rate, with 17% of noncitizen families utilizing Medicaid in 1999 (Fix and Passel 2002). The literature suggests several reasons for this steep decline in immigrant participation. One body of literature suggests that the decline in the immigrant participation was not necessarily due to the restrictions targeting them, but rather the booming economy of the 1990s. The late 1990s was a time of economic prosperity in the United States, with a low unemployment rate and an increase in the minimum wage. From 1994 to 1999, total cash assistance caseloads were reduced by half (Bell 2001). A 1997 Council of Economic Advisors report found that 31 to 44% of the decline in cash assistance caseloads between 1993 and 1996 could be attributed to the decreasing unemployment rate, while only 14 to 30% of the decline could be explained by the welfare restrictions imposed by PRWORA (Bell 2001; Moffitt 1999). Other studies took advantage of the variation in immigrant eligibility rules across states, explaining the drop in immigrant caseloads as a chilling effect (Fix and Passel 2002). In other words, even in states where immigrants were eligible, they chose not to apply for it (Borjas 2004). And finally, another line of research suggests that decreases in welfare use among immigrants was actually due to immigrants shifting their eligibility by becoming citizens. The 1990s was a period of increasing naturalization rates, and more than one million naturalizations occurred in 1996 (Van Hook 2003). Using Survey of Program Dynamics data, Van Hook (2003) found that a significant proportion of the decline in welfare caseloads among noncitizens was due to the naturalization of noncitizens. And yet, little is known about how Welfare Reform affected the children of immigrants whose families might have lost or been ineligible from receiving benefits as they enter young
10 adulthood. Among low-income immigrant pre-school children living in Chicago, those whose families lost access to benefits had worse health outcomes than their peers in immigrant families that retained access to benefits as well as children with native-born parents who stopped receiving welfare benefits (Kalil and Crosby 2010). Additionally, among the children of Latino immigrants, restricting welfare benefits increased the high school dropout rate (Condon et al. 2015). Bitler and Hoynes (2013) found that post-welfare Reform, low-income immigrant headed households with children were more likely to rely on earnings rather than the social safety net compared to native headed households with children. Additionally, after the Great Recession, Bitler and Hoynes (2013) found that immigrant headed households with children experienced greater increases in poverty than native headed households with children, but the social safety net provided less relief to low-income immigrant households than to native households. While providing valuable insight into the potential problems associated with the immigrant-based restrictions imposed by Welfare Reform, these studies are limited to small select samples (Condon et al. 2015; Kalil and Crosby 2010) and do not focus on young adults (Condon et al. 2015; Kalil and Crosby 2010; Bitler and Hoynes 2013). The current study addresses this shortcoming by examining the public assistance use among a cohort of young adults likely affected by Welfare Reform while living in the parental home. Public Assistance Use among the Children of Immigrants In the United States after Welfare Reform, it became increasingly difficult for immigrants to receive welfare benefits. The ruling that immigrants must have lived in the United States for at least five years before receiving public assistance or ten years for food stamps made it so immigrants had to join the labor force, often in low-paying entry level jobs (Reisel et al. 2012). Previous research has shown mixed results about immigrant welfare use levels compared to
11 natives, often depending on the welfare program and the income level of the family. For instance, among households with children, those headed by immigrants are equally as likely as those headed by natives to live in households that receive cash assistance. However, among lowincome households, they are less likely to receive cash assistance than those headed by natives (Chaudry and Fortuny 2010). There are also differences in welfare use among country of origin and mode of entry into the United States (i.e., between sponsored, refugee, and undocumented immigrants). The bulk of undocumented immigrants come from Mexico, Honduras, Guatemala, and El Salvador (Girard 2015). In 2012, over half of unauthorized immigrants to the United States were from Mexico (Passel, Cohn, Krogstad, and Garcia-Barrera 2014). Although results are mixed on the welfare use of undocumented immigrants compared to natives, Girard (2015) found that immigrants from the top four undocumented countries of origin had higher odds of receiving public assistance, as a whole, than natives. Perhaps unsurprisingly, immigrants who arrived as refugees also had higher odds of receiving public assistance than natives, but sponsored immigrants had lower odds of receiving public assistance than natives (Girard 2015). However, these results are based on adults aged 18-64, rather than being focused solely on young adult outcomes and assume mode of entry based on birthplace, leading to potential bias in the results. In addition to having higher levels of poverty, children of immigrants also have higher levels of food insecurity (Chaudry and Fortuny 2010). Whereas a third of native-born children grow up in households that receive some form of government assistance, almost half of children of immigrants do (Borjas 2011). There is some indication that this carries into adulthood, as groups by country of origin with the highest poverty and public assistance use rates in childhood have the highest poverty and public assistance use rates as young adults (Borjas 2011).
12 However, this finding is based on rates of poverty and public assistance at the height of the Great Recession and may not reflect the current cohort of young adult children of immigrants. This study will attempt to expand upon these findings and examine whether or not there is an intergenerational transmission of welfare use among the children of immigrants. Intergenerational Transmission of Welfare Use There is a large literature on the intergenerational transmission of welfare use, although few papers focus on immigrants specifically. For example, using the National Longitudinal Survey of Youth 1979 (NLSY79), Antel (1992) has shown that maternal welfare receipt is positively associated with daughter s receipt later in life. There is much debate as to whether the positive relationship between parental and child welfare use is due to a transmitted welfare dependency (Butcher and Hu 2000; Borjas and Sueyoshi 1997; Levine and Zimmerman 1996). Levine and Zimmerman (1996) argue that in order to be eligible for welfare, you have to fall below a certain income level. In their studies of mothers and daughters welfare use, Levine and Zimmerman (1996) found that over 75% of the correlation between mothers and daughters welfare use could be attributed to their economic status. Therefore, they suggest that the intergenerational transmission of welfare use is due to the transmission of familial characteristics (such as low education levels) and income, suggesting a cycle of poverty over a cycle of welfare dependence. Borjas and Sueyoshi (1997) found that ethnic differences in welfare use persisted from one generation to the next. Using data from the 1980 Census for the parent generation and data from the NLSY79 for the child generation, Borjas and Sueyoshi (1997) found among all young adults, the probability of welfare receipt increased by 23% when their parents received welfare. Higher parental education reduced the probability of welfare receipt among the child generation.
13 Hispanic and black groups had high participation rates among both the parent and child generation, and therefore had a high intergenerational transmission rate. However, even when looking at only non-hispanic white groups, a significant correlation between parental and child welfare receipt existed (Borjas and Sueyoshi 1997). Limited literature has examined the intergenerational transmission of welfare use among immigrants. However, (Butcher and Hu (2000) have shown that there is a positive association between parental and child receipt, but it is weak. Similar to Borjas and Sueyoshi (1997), Butcher and Hu (2000) found that the variations in welfare use vary drastically by country of origin. Cuba and Mexico have high welfare receipt rates in the parent generations, but below average welfare receipt rates in the child generation. Butcher and Hu (2000) came to the same conclusion as Levine and Zimmerman (1996), finding that children of immigrants whose parents are poor are more likely to be poor themselves, supporting the notion of a cycle of poverty rather than a cycle of welfare dependence. These cohort studies, using data from the 1980s and mid- 1990s, do not capture the potential impact of welfare based restrictions on the propensity to use public assistance in later generations. These findings are expanded upon in the current study, using the mean grouping estimation strategy employed by Borjas and Sueyoshi (1997), Butcher and Hu (2000), and Balistreri (2010). More recent research on the intergenerational transmission of welfare use among immigrant generations has examined its link to educational attainment among the child generation. Having recently received welfare benefits is associated with 63% reduction in the odds of high school graduation for the children of natives, but only a 32% reduction in odds for the children of immigrants. Similarly, recent welfare receipt is associated with a reduction in odds of college enrollment of 59% for children of natives but only 12% for the children of
14 immigrants (Balistreri 2010). However, Balistreri (2010) found that once country of origin was controlled for, children of immigrants, except those from Mexico and Central America, had higher odds of high school graduation than the children of natives did when welfare receipt was held constant. These results indicate that welfare receipt in the childhood home could provide immigrant parents with more resources to invest in their children, helping the children of immigrants to achieve better educational and socioeconomic statuses later in life than they might have otherwise (Balistreri 2010). However, these data are based on outcomes of the young adult children of immigrants prior to the immigration-based restrictions of Welfare Reform.
15 THE CURRENT STUDY Given the existing gaps in the literature, the current study will address three research questions. The first research question is, Among the young adult population, are the children of immigrants more or less likely to be low-income than the children of natives? That is, are they disproportionately represented among the low-income population that would be the most likely to not only be eligible for public assistance but need it? Given that the young children of immigrants are more likely to live in poverty than their peers with native-born parents (Woods et al. 2016) and are more likely to be food insecure (Chaudry and Fortuny 2010), it may be that the young adult children of immigrants age into poverty at higher rates than the children of natives. Indeed, prior research has found that although the poverty rate declined for all households between 1996 and 2000, the poverty rate increased by 5% at the height of the Great Recession (between 2007 and 2009) for immigrant-headed households with children, but remained stable for native-headed households with children (Borjas 2011). While the safety net helped dampen the effects of the Great Recession for native headed families with children, it did not provide much relief for immigrant headed families with children (Bitler and Hoynes 2013).This suggests that the Great Recession had a greater negative economic impact on immigrant families than native families. On the other hand, despite many hardships faced by immigrant families, for most immigrant groups, the socioeconomic attainments of the children of immigrants not only exceed their parents but also their native-born peers (Crul et al. 2012; Hofferth and Moon 2016; Tran and Valdez 2015). For example, with some exceptions, previous research has found that the children of immigrants achieve higher education levels than their immigrant parents and nativeborn peers (Crul et al. 2012; Hofferth and Moon 2016; Tran and Valdez 2015). Additionally,
16 previous research has found that the young adult children of immigrants are less likely to work in manufacturing than their immigrant parents, and have similar professions and labor force participation as the young adult children of natives (Kasinitz et al. 2008). Based on their high socioeconomic achievements, it may be that the young adult children of immigrants will be less likely to be low income than their peers with native born parents. This question will be answered using the 2012-2016 CPS young adult sample. The second research question investigated in this study is, Among the low-income young adult population, are the children of immigrants more or less likely to utilize public assistance? Previous research on welfare use among the children of immigrants has been mixed, and is focused mainly during their childhood. Nearly half of children of immigrants grow up in households that receive some form of public assistance, compared to one-third of children of natives (Borjas 2011). Recent studies focusing on low-income households have found that as a whole, immigrants are less likely to receive public assistance such as cash assistance and food stamps. However, more research is needed to determine whether these findings persist among the children of immigrants into adulthood. Particularly, this study is unique in that it focuses on children of immigrants who lived in the parental home when Welfare Reform occurred, but also were in young adulthood post- the Great Recession. The children of immigrants sample used in this study contains the children of immigrants who have spent either most or all of their lives in the United States, therefore it could be that they will have similar levels of welfare receipt as their peers with native born parents. This research question will be answered using the 2012-2016 CPS young adult sample. The third research question is, Is there an intergenerational transmission of welfare use between the immigrant parent generation and the current generation of young adult children of
17 immigrants? Previous research has found evidence of an intergenerational correlation in welfare receipt (Antel 1992; Butcher and Hu 2000; Borjas and Sueyoshi 1997; Levine and Zimmerman 1996). However, much of the intergenerational correlation is accounted for by differences in ethnic groups, with certain ethnic groups having higher levels of poverty and public assistance use than others (Butcher and Hu 2000; Borjas and Sueyoshi 1997). Particularly, groups with low parental education levels, high concentrations in poverty stricken neighborhoods, and parental lack of social capital could be at a disadvantage when entering adulthood, making it difficult for them to have upward social and economic mobility (Crul et al. 2012). While previous research has found that parental welfare receipt could provide immigrant parents with more resources to invest in their children s future, it could be that families who were cut off from welfare receipt after Welfare Reform were unable to invest as much in their children as their peers who retained access to benefits, making them more likely to be lowincome and need public assistance. Additionally, previous research has suggested that this can be explained by a cycle of poverty, and the parent to child transmission of socioeconomic characteristics rather than the transmission of welfare dependence (Levine and Zimmerman 1996; Butcher and Hu 2000). This research question will be answered using data from the 2001 ACS parent sample and the 2012-2016 CPS young adult sample.
18 DATA AND METHODS To address the research questions of the current study, two analytic samples will be created a young adult sample and an immigrant parent sample. The first is a sample of the young adult children of immigrants and the young adult children of natives from the 2012-2016 Current Population Survey Annual Social and Economic Supplement (ASEC). The young adult age group has been used in previous studies examining the intergenerational transmission of welfare use (Borjas and Sueyoshi 1997; Butcher and Hu 2000) and how parental program participation affects children of immigrants socioeconomic attainments (Balistreri 2010). Young adults aged 25-29 are an ideal age group for this study, since they are starting to form their own households, have typically completed their education, and are entering the labor force. The second analytic sample includes the immigrant parent generation derived from the 2001 American Community Survey. The 2001 American Community Survey was chosen due to its large sample size, public assistance use measures, and the fact that the data were collected five years after the enactment of Welfare Reform, when the nativity-based restrictions would have been well-established. Each are described below. Current Population Survey The Current Population Survey (CPS) is a national monthly survey that has been collected since the late 1930s, with the United States Census Bureau collecting the data since 1942. Currently, the CPS is a joint survey between the Census Bureau and the Bureau of Labor Statistics. The survey aims to provide a current profile of the labor force, with much of the focus on the employment experience of its participants in the week prior to the survey.
19 In addition to the monthly surveys, the CPS also includes the Annual Social and Economic Supplement (ASEC) every March, which provides additional social, demographic, and economic variables. Due to its broad range of variables and large sample size, the ASEC is well suited for the current study. Respondents in the CPS are in the sample for eight months total: four months at the beginning, an eight month break, and then four months before completing the survey and exiting the sample. Due to the nature of the on four, off eight, on four survey design, it is possible that some respondents are in adjacent years of the ASEC. For this project, data from the 2012-2016 CPS ASEC will be combined. To account for respondents in multiple years, only respondents who were in their first four months of completing the survey in 2012, 2013, 2014, and 2015 will be included. All respondents from 2016 will be included. Each ASEC contains responses from roughly 98,000 households from around the United States. There are 1,328 counties included in the sample, with each state and the District of Columbia represented. The ASEC also includes a sample of 450 Armed Forces members living with civilians on or off base, as well as an over-sample of the Hispanic population. The CPS ASEC is an ideal data set for the current project, because it includes information about the respondent s immigration history (i.e., country of birth, year of arrival, citizenship status, and parents country of birth), as well as educational attainment and detailed information regarding their employment status. In addition, the ASEC also provides information on multiple types of welfare and government assistance receipt at both the individual and household level. Initially, the 2012-2016 CPS ASEC sample contained 584,477 respondents. The total sample of combined years included 35,622 individual respondents ages 25 to 29 years. Foreignborn respondents who came to the United States over the age of 12 were removed from the sample, resulting in a sample size of 31,051. Of that sample, 24,976 are children of natives, and
20 6,075 are children of immigrants (children of immigrants who are either native born, or came to the United States before the age of 13). Low-Income Sample To evaluate the public assistance use among those who are most likely to be eligible, a low-income sample was created of respondents 185% below the poverty line the typical cut point for public assistance programs. The low-income sample is taken directly from the young adult CPS sample, and includes 1,802 children of immigrants and 7,861 children of natives, for a total of 9,663 respondents. American Community Survey In the current study, the 2001 American Community Survey (ACS) is utilized to study intergenerational transmission of welfare receipt. The ACS is an annual survey conducted by the United States Census Bureau. It was designed to replace the long-form decennial census and aims to provide an up-to-date demographic portrait of the United States. The 2001 ACS is designed to be a 1 in 232 nationally-representative sample of the non-institutionalized population of the United States. Due to the large sample size and information on birthplace and public assistance use, the ACS is ideal to use to gather information on the immigrant parent generation. For the immigrant parent sample, households were targeted that had children ages 11 to 18 present to roughly match the birth cohort of the young adult sample. There were 84,055 households with children in the target age range. All households with a foreign-born household head or spouse were identified, and country of origin assigned. For households where both the head and spouse was foreign born, the country of origin was assigned based on the country of birth of the female respondent. Countries of origin with fewer than 40 households present in the
21 sample were removed in order to have a large enough sample for adequate representation. The final sample consists of 11,880 households from 49 different countries where either the head or spouse (or both) are foreign born. Dependent Variables Cash Assistance is a dichotomous variable indicating whether the respondent received cash assistance in the form of Supplemental Security Income (SSI) or any other welfare in the form of cash in the past year. In the current analysis, not receiving cash assistance will be the reference category. In the CPS, cash assistance is measured at the individual-level, whereas in the ACS, cash assistance is measured at the household-level. A household-level cash assistance variable was created for the CPS in the intergenerational analysis, to increase comparability between the two samples. Supplemental Nutrition Assistance Program (SNAP), also known as food stamps, is a dichotomous household-level variable, coded 1 if the respondent lived in a household that received SNAP in the past year and zero if the respondent s household did not receive food stamps in the past year. Food stamp receipt is an individual-level variable in the ACS, but a household-level variable was created for the intergenerational analysis to match the householdlevel variable in the CPS. Medicaid Receipt is an individual-level dichotomous variable coded 1 if Medicaid covered the respondent at any time in the past year and 0 if the respondent was not covered by Medicaid at any time in the past year. Medicaid receipt is only available in the CPS sample. Any Public Assistance Use is a dichotomous variable coded 1 if the respondent (for cash assistance and Medicaid) or anyone in their household (for food stamps) received cash
22 assistance, food stamps, or Medicaid benefits. This variable was only created for the CPS sample and is used in the logistic regression analyses. High School Graduate is an individual-level dichotomous variable indicating whether or not the respondent graduated from high school. Since the age category in this analysis is from 25-29, the respondents should be at the age where they would have completed high school. The reference category is those who did not graduate from high school. This variable was created for both data sets and is used only in the intergenerational transmission analyses. Ever Enroll in College is an individual-level dichotomous variable indicating whether the respondent enrolled in college at any point in time. The CPS ASEC contains information on the educational attainment of the respondent. If the respondent completed some college but no degree or obtained an Associate s degree or higher, the respondent is considered to have enrolled in college. It only includes respondents who graduated from high school. For my analysis, those who did not enroll in college will be the reference category. This variable was only created for the CPS sample for the intergenerational analyses. Independent Variable Throughout the logistic regression analyses using the CPS ASEC data, the key independent variable is whether the respondent is the child of immigrants. This is a dichotomous variable, with the children of immigrants coded as a 1 and children of natives as the reference category. Control Variables The control variables for the logistic regression analyses, which only use 2012-2016 CPS data, are as follows:
23 Race is a categorical variable, including white (reference), black, Hispanic, Asian, or member of another racial category. or female. Gender is a dichotomous variable, indicating whether the respondent is male (reference) Relationship to Household Head is a dichotomous variable indicating whether the respondent lives in an independent household, and is either the household head, spouse of the household head, or has another relationship to the head (reference) or lives in the parental home, and is either the child or grandchild of the household head. Educational Attainment is a categorical variable, including less than a high school degree, high school diploma or equivalent (reference), some college, or Bachelor s degree or higher. Marital Status is a categorical variable, including currently married (reference), previously married, and never married. Region of Origin is a categorical variable indicating the region of the world that the respondent was born in if they were born outside of the United States, or the region of the world that their parents were born in if they are native born with foreign born parents. If both parents were born in different regions of the world, the mother s birthplace was chosen. Region of origin includes native born (reference), Central America, Caribbean, South America, Europe/Canada, Asia/Oceania, and Africa. Europe and Canada were combined into one category due to the small number of children of Canadian immigrants in the low-income sample. In addition, because I am using pooled data from 2012-2016, I will include a set of dummy variables indicating survey years to account for period effects. I will also control for
24 state fixed-effects, since there is variation in welfare generosity by state (Potochnick 2016; Zimmermann and Tumlin 1999). The intergenerational analyses will include controls for age and age squared for the parent generation, as well as the percentage of high school graduates for the parent generation by country of origin.
25 ANALYSIS PLAN To address the research questions in the current study, I conduct three sets of analyses to examine the differences between the young adult children of immigrants versus their native-born peers in terms of low-income status and welfare use. First, I present a descriptive analysis with mean welfare use, and key demographic and socioeconomic characteristics for the children of immigrants and the children of natives (Table 1). Additional analysis presents these statistics for the low-income subsample (Table 2). This descriptive analysis examines the overall difference in welfare use and achievements between children of immigrants and the children of natives. Logistic regression predicting low-income status (Table 3) addresses the first research question, Among the young adult population, are the children of immigrants more or less likely to be low-income? This analysis describes who is in the low-income sample, and determines whether the children of immigrants are similarly likely to be low-income as the children of natives, controlling for socioeconomic and demographic characteristics. To address the second research question, Among the low-income young adult population, are the children of immigrants more or less likely to utilize public assistance? Here, the sample is limited to households at or below 185% of the federal poverty line using the lowincome sample, since this is the typical cut point for public assistance eligibility. This analysis assesses the likelihood of program participation for cash assistance, food stamp receipt, Medicaid use, and any public assistance receipt among low-income households by demographic characteristics and parental nativity status (Tables 4, 5, 6, and 7). To explore whether there is any evidence of an intergenerational transmission of welfare use, I first present descriptive statistics of the 2001 ACS parent sample and the 2012-2016 CPS
26 child sample to compare their public assistance use, poverty levels, and socioeconomic attainments (Table 8). Finally, I conduct a grouping estimation procedure similar to Borjas and Sueyoshi (1997), Butcher and Hu (2000) and Balistreri (2006). The purpose of this analysis is to provide some descriptive evidence on the intergenerational correlations of welfare use between the immigrant parent and young adult children of immigrant generations in the United States (Table 9). For this analysis, a grouping estimation strategy is used that creates aggregate level country of origin groups instead of micro level individual data. The equation will be specified as follows: Ϛjk 2 = α + βϛjk 1 + εj where Ϛjk 2 and Ϛjk 1 are the mean receipt rates of receipt for program k by the jth origin group of the native-born children of immigrants in the 2012-2016 Current Population Survey and their prospective immigrant parents in the 2001 ACS, respectively. The purpose is to estimate the coefficient β, which represents the percentage point increase in the mean rate of receipt for program k among the young adult children of immigrants that is associated with a one percentage point increase in the parents generations mean rate of receipt for program k. A coefficient of zero indicates that the immigrant parents welfare receipt has no impact on the subsequent use of the child generation, whereas a coefficient of one suggests that the parents welfare use will be replicated in the child generation. Only those who are children of immigrants are included in this analysis; children of natives are excluded. Each group (immigrant parents and the children of immigrants) is divided into the same 49 country of origin groups. First, the head of household age-adjusted mean receipt rates for the parent cohort and the child cohort are estimated for cash assistance and food stamp receipt. The child cohort rate of use is regressed on the parent cohort rate of use for each