Educational Effects of Banning Access to In-State Resident Tuition for Unauthorized Immigrant Students

Abstract

This research examines the effects of state laws banning access to in-state resident tuition for unauthorized immigrant students in the United States. These laws were implemented between 2005 and 2012. We evaluate the policy effects on (a) college enrollment, (b) school dropout rates of unauthorized immigrants, and (c) the enrollment of U.S. citizens in higher education. Multivariate triple-differences models are used. We find significant negative effects on the college attendance rates of unauthorized immigrants. Policies have primarily affected recent high school graduates. With regard to dropping out of school, we find no evidence of dynamic effects. Nor do we find evidence of benefits in college attendance for non-Hispanic, Hispanic, or Mexican naturalized citizens.

Keywords

unauthorized immigrants in-state tuition college access state policy evaluation natural quasi-experiment

The purpose of this research is to examine the effects of banning access to in-state resident tuition (ISRT) and other state financial benefits such as scholarships and grants for unauthorized immigrant students (UIS).¹ Public policy providing or restricting access to ISRT and other higher education benefits for this group has been developed within a diverse and complex environment in the United States. Influenced by federal laws, and ultimately defined by policymakers at the state level, legislation on this issue has been the product of a wide array of social, economic, and political conditions in each state. Not surprisingly, because of the different conditions among states as well as the relationship between the national government and the states, a range of policies have been adopted. Some states prohibit access to public postsecondary institutions for UIS, whereas others provide not only ISRT but also private and state-funded grants and scholarships for this population. The ISRT theme has reached the federal and state government agendas on multiple occasions for more than a decade (Frum, 2007; National Conference of State Legislatures, 2011, 2014b; Olivas, 2004, 2009). Although there are no federal laws that explicitly prohibit the admission of UIS to higher education, federal legislation makes these individuals ineligible for federal financial aid and restricts their access to state financial aid. Therefore, under federal provisions, UIS wanting to pursue postsecondary studies have to pay out-of-state tuition no matter how long they have resided in the state, and have no access to any state financial help unless the same state benefits are given to every other citizen, even nonstate residents (Illegal Immigration Reform and Immigration Responsibility Act [IIRIRA], 1996; Olivas, 2004; Personal Responsibility and Work Opportunity Act [PRWORA], 1996). In an extreme scenario, states or higher education institutions can declare admission of UIS to higher education public institutions illegal. This policy contrasts with the treatment that must be provided by public institutions in Grades K through 12. In the case of K–12 education, unauthorized immigrants have a constitutional right to receive instruction at public institutions in the United States (Plyler v. Doe, 1982). No such constitutional right exists for postsecondary education.

At the federal level, the issue was first framed within two bills dealing with broad topics: immigration and noncitizen eligibility for federal public assistance programs, in 1996. The IIRIRA indicates that states are not able to provide a higher education benefit based on state residency to unauthorized immigrants unless the same benefit is provided to all eligible U.S. citizens, regardless of state residency (IIRIRA, 1996). In other words, “State A cannot give any more consideration to an undocumented student than it can give to a nonresident student from State B” (Olivas, 2004, p. 453). In addition, through the PRWORA, Section 401, Congress provided access to federal public benefits, including postsecondary education financial assistance, exclusively to those “qualified aliens” legally present in the country (PRWORA, 1996).

In 2008, regarding the access of UIS to higher education, the U.S. Department of Homeland Security, Immigration and Customs Enforcement clarified that the enrollment of UIS in public postsecondary institutions does not violate federal law; it is a decision of the states, and a decision of the institutions if no state law exists (Olivas, 2009; Russell, 2011). Under this scenario, the possibility of unauthorized immigrants having access to ISRT and other state educational benefits has depended on the state of residence and its interpretation of federal immigration legislation as well as the state regulation of the higher education system (Olivas, 2004, 2009).

Unauthorized immigrants’ access to higher education benefits funded with state resources varies across states. Appendix Table A1 presents the states that have advanced any type of measure on this topic. As of Summer 2014, 16 states—California, Colorado, Connecticut, Florida, Illinois, Kansas, Maryland, Minnesota, Nebraska, New Jersey, New Mexico, New York, Oregon, Texas, Utah, and Washington—allow access to ISRT for unauthorized immigrants through state legislation. Five of those states—California, Minnesota, New Mexico, Texas, and Washington—also allow unauthorized students to receive state financial aid in the form of scholarships and grants. Four states—Hawaii, Michigan, Oklahoma, and Rhode Island—through Board of Regents’ decisions have also opened the possibility for unauthorized immigrants to pay in-state tuition rates. In contrast, six states—Arizona, Colorado, Georgia, Indiana, Montana, and Ohio—have enacted laws banning UIS from receiving ISRT and any other type of state financial aid. Two states, Arizona and Montana, approved ISRT prohibition by referendum, but the measure in Montana was later overturned by a District Court.² Finally, the most extreme measures have been taken by Alabama and South Carolina where the enrollment of unauthorized immigrants in state higher education institutions is prohibited (Education Commission of the States, 2014; Institute of Higher Education Law and Governance, 2014b; National Conference of State Legislatures, 2014b).

The discussion about the eligibility of UIS to pay ISRT is important because of the number of people potentially affected by the measure. According to the most recent estimate available from the Pew Hispanic Center, 11.7 million unauthorized immigrants were living in the United States as of March 2012, a half million people less than in 2007 (Passel, Cohn, & Gonzalez-Barrera, 2013). Among this population is a group identified as the “1.5 immigrant generation,”³ which consists of foreign-born children who were brought by their parents to the United States and have grown up in the country. An estimation of the impact of this group of immigrants on the present and future demand for higher education in the United States, projected by the Migration Policy Institute, showed that in 2012, 140,000 unauthorized immigrants were enrolled in college; 390,000 were high school graduates or had a General Education Development; and 800,000 were enrolled in K–12 institutions (Batalova & Mittelstadt, 2012). Although the percentage for U.S. recent high school completers was 66.2 in 2012 (National Center for Educational Statistics, Digest of Education Statistics, 2013), it is estimated that of the 65,000 UIS who annually graduate from high school, only about 5% to 10% enroll in postsecondary education (Russell, 2011).

Research on the empirical effects of state legislation that defines the access to ISRT and other higher education benefits for UIS is scarce. Existing research essentially focuses on analyzing the effects of laws providing access to ISRT, rather than on those laws that forbid it. Basically, both types of policy result in changes of the same magnitude, but opposite direction, in the price that UIS have to pay for higher education. If UIS value gains and losses in tuition equally, it would be expected that the effect of banning access to ISRT would be of the same magnitude, but in the opposite direction, of policies providing that benefit. Prospect theory, however, maintains that people value gains and losses differently (Kahneman & Tversky, 1979), and consequently, it may be the case that the effect of policies allowing and banning ISRT for UIS produce different results not just in terms of the direction of the impact but also in terms of magnitude. This research examines the effects of the latter group of policies in five states: Arizona, Colorado, Georgia, Indiana, and Ohio. We estimate the overall effect of policies denying ISRT implemented between 2005 and 2012. Three potential effects are evaluated. First, we estimate the policy effects on the enrollment of UIS in higher education. Second, considering the potential dynamic effects of policies banning access to ISRT for UIS, we evaluate the effects of ISRT policies on the school dropout behavior of UIS. The dynamic effects of the policy are based on the idea that the lack of real opportunities to attend higher education, among other factors, demotivates UIS from attending secondary schools, prompting them to drop out of school (Abrego, 2006; Chin & Juhn, 2011; Gonzales, 2011; Suárez-Orozco, Yoshikawa, Teranishi, & Suárez-Orozco, 2011). Finally, we estimate the effects of banning ISRT access for UIS on the enrollment of citizens and legal residents in higher education institutions. We seek to answer the following research questions:

Research Question 1: How did prohibition of access to ISRT and state financial aid to UIS affect their college participation rates during 2005 to 2012?

Research Question 2: How did prohibition of access to ISRT and state financial aid to UIS affect their high school dropout behavior during 2005 to 2012?

Research Question 3: How did prohibition of access to ISRT and state financial aid to UIS affect college participation rates of U.S. citizens and legal residents during 2005 to 2012?

This article consists of five sections, including this introduction. Section “Literature Review and Theoretical Framework” presents a review of scholarship on the effects of access to ISRT for unauthorized immigrants, the theoretical framework, and the hypotheses. Section “Research Design” describes the research design, including data sources, sample, analytic strategy, and the method. Descriptive analysis, findings, and robustness tests are presented in section “Findings.” The last section presents the conclusions and discussion, research limitations, and policy recommendations.

Literature Review and Theoretical Framework

Scholarship on the Effects of ISRT for UIS

Methodologically, a distinguishing characteristic of the studies analyzing the effects of giving access to in-state tuition rates for UIS has been merely to estimate the population targeted by the policy (i.e., UIS). Because no U.S. government agency directly registers undocumented migrants (Passel, 2005), studies that have measured the overall effects of the policy on a group of states or a single state have had to work with samples of Hispanics or Mexicans who are highly likely to be undocumented using the Current Population Survey (CPS; Amuedo-Dorantes & Sparber, 2014; Darolia & Potochnick, 2015; Flores, 2010a, 2010b; Flores & Chapa, 2009; Kaushal, 2008; Potochnick, 2014) or the American Community Survey (ACS; Chin & Juhn, 2011). However, studies that examined the policy effects at the institutional level identified UIS through the use of specific administrative databases that report the students affected by the policy (Conger & Turner, 2015; Dickson & Pender, 2013).

The main conclusion in terms of the effects of policies giving access to in-state tuition rates for UIS is that this type of policy has produced positive and statistically significant effects on the educational outcomes of students highly likely to be undocumented and no harmful effects on legal immigrants or U.S. citizens. Kaushal (2008), using multivariate regression models with robust standard errors clustered on state-year, found that access to ISRT rates for Mexican young adults, who were highly likely to be both undocumented and beneficiaries of the policies, increased the college enrollment and education of this population. In short, using a sample of young Mexican adults, the policy was associated with a 2.5 percentage point increase in college enrollment, a 3.4 percentage point increase in the proportion with at least a high school diploma, a 3.7 percentage point increase in the proportion with at least some college education, and a 1.3 percentage point increase in the proportion with a college degree. In addition, the study showed no evidence of adverse effects of the policy on the educational outcomes of the U.S.-born college age population, and positive effects on the college enrollment of U.S. citizens of Mexican parentage. Flores (2010b), focusing on Latinos and employing difference-in-differences analysis, found positive and statistically significant effects of the tuition policies on the odds of UIS being enrolled in an institution of higher education. She found that the odds of college enrollment among Latino individuals who are highly likely to be undocumented living in those states with in-state tuition for UIS were 54% higher after the implementation of the policy. Amuedo-Dorantes and Sparber (2014), using a sample of Mexican unauthorized immigrants from the complete CPS, estimated an approximately 4 percentage point increase in college enrollment of this population. Unlike Kaushal (2008) and Flores (2010b), they found adverse policy effects on foreign-born non-Mexican Hispanics and Hispanic foreign-born citizens. Darolia and Potochnick (2015) found that in-state tuition results in positive effects of approximately 2 percentage point on college enrollment of Latino foreign-born noncitizens (FBNCs); however, their results are only marginally significant. Potochnick (2014), examining the effects on a different indicator, but also employing difference-in-differences strategy, found that ISRT policies allowing access to ISRT for UIS caused a statistically significant reduction of 8 percentage points in the proportion of young Mexican (ages 16–19) unauthorized immigrants that drop out of high school.

Two studies looking at state-level effects are Flores (2010a) and Flores and Chapa (2009). Similar to Flores (2010b), both studies used logistic regression to estimate differences-in-differences models. Flores (2010a) focused on the Texas case whereas Flores and Chapa (2009) focused on the group of all states that implemented the policy before 2006. Analyzing the period 1998 to 2004, Flores (2010a) found that chances of attending college education among Latino FBNC students, which were highly likely to be undocumented, were higher after Texas adopted the policy. The strongest effect of policy implementation was on older high school graduates (ages 21–24), whose odds were 84% higher. The study also found that the population targeted by the policy tended to enroll more in community colleges rather than in 4-year institutions. There were increasing yearly effects of the policy from 2001 to 2003, but no effect for 2004, and the effect of the policy on college enrollment odds was captured particularly by the target population, that is, Latino UIS, and not by all Latinos or U.S.-citizen Latinos. Meanwhile, Flores and Chapa (2009) claimed that in-state tuition policies can differ among states depending on state-level settlement migration patterns. Therefore, the authors sought to study the effect of this policy controlling for Latino migration settlement patterns. The findings suggested that the policies implemented in those states seen as traditional settlements of Latino population had a significant impact on the enrollment of Latino foreign-born students. Their odds of college enrollment were 69% higher than those of their peers in states with similar Latino migration patterns, but without the policy. However, states experiencing new Latino migration settlements seemed to have no significant effect on the enrollment of Latino foreign-born students after the implementation of the policy, as compared with similar states without the policy. Finally, after the implementation of the policy, the target population in the traditional Latino destination region had higher chances of enrolling than did the same population in southern states without an ISRT policy. Thus, their odds of college enrollment were 79% higher.

Texas in 2001 and New York in 2002 were two of the first states passing laws giving access to ISRT rates for UIS; using administrative data and difference-in-differences identification strategy, Dickson and Pender (2013) and Conger and Turner (2015) examined the effects of the policy in five public universities in Texas and New York City’s public university system, respectively. The first study found that the reduction in the education cost generated by the law implemented in 2001 in Texas produced (a) large and positive (11 and 18 percentage point increase) significant effects on the probability of enrollment of noncitizens accepted at two public institutions that already served a large number of Hispanics, (b) no significant effects on the enrollment probability of noncitizens accepted to state flagship universities, and (c) significant but opposite results (+15 and −2 percentage points) at two universities when the sample was limited to Hispanic students.

In the New York City study, Conger and Turner (2015) estimated the effects of a temporary suspension of ISRT benefit on the attainment of enrolled unauthorized immigrant college students. They found that reenrollment of senior UIS during the semester of suspension fell by 8% whereas that of UIS college students had no effect; however, this group had a reduction in the credits earned by about 8%. With regard to long-term effects, the authors estimated that degree receipt after 8 years of entry fell 22% among senior college students and no effects on community colleges students.

The gap in the study of the educational effects of in-state tuition policies is a lack of analysis of state laws that ban the access to ISRT for UIS. Despite the significant number of unauthorized immigrants living in these states, the effects of in-state tuition laws enacted in Arizona and Colorado in 2006, Georgia in 2008, and Indiana and Ohio in 2011 have not been studied.⁴ Unlike the laws enacted in South Carolina in 2008 and Alabama in 2011, which prohibited UIS from enrolling in state’s colleges or universities, the laws merely banning access to in-state tuition for UIS leave the gates of higher education open for this population, but at a higher cost. These costs can make higher education prohibitive for many UIS. The educational effects of banning access to ISRT for UIS on this population as well as on naturalized and native citizens are unknown.

Conger and Turner (2015) is the only work that estimated the effects of a policy denying access to ISRT on UIS. However, they only consider students enrolled in college, so their results are conditional on previous enrollment. In addition, the policy analyzed by them was a temporal measure that lasted one academic term. Our research adds to the current scholarship on this topic by measuring the effects of persistent policies banning ISRT on both unauthorized immigrants previously enrolled in a higher education institution as well as on those who were not previously enrolled. Also, the demand for higher education by UIS may respond differently to increases and decreases in the price, which makes it worthwhile to determine the impacts of restrictive ISRT policies and compare them with the results of previous works on the opposite policies. We also estimate the effects of banning ISRT for UIS on dropping out of school among this group of people, and the enrollment effects on U.S. citizens.

Theoretical Framework

Human capital theory indicates that investments in people produce economic benefits to individuals as well as society (Sweetland, 1996). Since the pioneering works of this theory during the early 1960s, education has been considered one of the most worthwhile investments in human capital along with health and job training (Becker, 1962, 1993; Shultz, 1963). According to human capital theory, these investments enhance individuals’ “mental and physical abilities,” which at the same time enhance their productivity. In a market economy, the productivity differentials of alternative human capital investment decisions result in lifetime earnings differentials where higher investments are rewarded through higher earnings (Becker, 1962).

In the study of the acquisition of education, the human capital model has been used extensively in both theoretical and empirical works (DesJardins & Toutkoushian, 2005). The human capital model assumes that students are rational decision makers who compare all monetary and nonmonetary expected direct and indirect costs with all the monetary and nonmonetary expected benefits associated with educational alternatives (DesJardins & Toutkoushian, 2005; Paulsen & Toutkoushian, 2008). Authors studying the educational decisions of unauthorized immigrants have recognized that in addition to the usual factors involved in the college decisions of traditional groups, such as those presented above, unauthorized immigrants face additional conditions that must be considered. Incomplete information and uncertain conditions faced by undocumented students make the decision to invest in college education even more complex. Some of the uncertain conditions faced by unauthorized immigrants include the lack of means to cover the payment of tuition and fees or to secure access to financial aid, the deportation risk, and the fact that after finishing the educational program, an undocumented student is not allowed to be legally hired in the labor market (Chin & Juhn, 2011; Flores, 2010a, 2010b; Kaushal, 2008).

In the context of the human capital model, the analysis of the effect of state policies banning access to ISRT for UIS exhibits two potential scenarios depending on the prepolicy conditions faced by this group in terms of access to in-state tuition rates. The first scenario assumes that UIS did have access to ISRT before banning policies were implemented. The Illegal Immigration Reform and Immigrant Responsibility Act of 1996 banned states from providing public benefits to unauthorized immigrants based on residence criteria, but Stevenson (2004) claims that the practice of a “longstanding policy of ‘don’t ask, don’t tell’” allowed UIS to have access to ISRT at public colleges even before some states started to enact policies regulating the issue (pp. 576–577). The administrative arrangement present in some states without statutory law has left the decision of access to ISRT for UIS to each institution on a case-by-case basis (Bell Policy Center, 2005, April) or to bureaucratic rule making (Thangasamy, 2007). Under these conditions, the implementation of policies banning access to ISRT for UIS clearly would mean an increase in the monetary costs of postsecondary education, from paying ISRT to paying out-of-state or international student rates. This would make it even harder for unauthorized immigrants to pursue higher education.

However, the second scenario assumes that prior to the enactment of state legislation banning access to ISRT for UIS, this group of people had no access to the education benefit. This would reflect the fact that states interpret federal legislation as prohibiting them from providing in-state tuition to undocumented students (as was the case across the country until late 1990s; Romero, 2002). Even though California and Texas adopted progressive legislation in favor of UIS in 2001, there are still states that deny access to ISRT based exclusively on federal legislation. Under this scenario, it might seem that the implementation of policies banning access to ISRT for UIS would not affect human capital investment decisions given that no change in direct monetary costs would take place. The new policy would merely confirm the previous conditions under which UIS have had to pay out-of-state tuition rates. However, the new policy actually increases both the nonmonetary costs as well as the uncertainty associated with the possibility of acquiring a college education. For instance, the policy can be perceived by UIS as a signal of animadversion toward them on campuses, as an increase in the risk of deportation, or as a form of labeling and discrimination.

Therefore, according to the human capital model, irrespective of which of the two scenarios in terms of prepolicy access to ISRT for UIS is considered, the ultimate effect of implementing state policies banning that benefit is an increase of the expected costs associated with preparing for, enrolling in, persisting in, and graduating from college. Thus, it is projected that the prohibition of giving access to ISRT for UIS increases the expected monetary and nonmonetary costs of attending higher education for this group, thus discouraging them from seeking postsecondary education. Also, negative policy side effects would be expected on the enrollment of UIS in secondary education because the ban would work as a disincentive for UIS to graduate from high school given the harder conditions they have to face to keep advancing in the education pipeline. However, because the policy has no effects on the monetary costs of higher education for citizens and naturalized citizens, it is expected that the effect of the policy on the higher education enrollment of this group is negligible.

Based on the literature review and these theoretical considerations, three hypotheses are proposed:

Hypothesis 1: There are negative effects on the college participation rates of UIS in the group of policy states as result of the prohibition of access to ISRT and state financial aid to this population.

Hypothesis 2: There are positive effects on the high school dropout rate of UIS in the group of policy states as result of the prohibition of access to ISRT and state financial aid to this population.

Hypothesis 3: There are no effects on the college participation rate of U.S citizens and legal resident students as result of the prohibition of access to ISRT and state financial aid to UIS in the group of policy states.

Research Design

Drawing primarily on research designs from studies that estimate effects across states (Chin & Juhn, 2011; Flores, 2010b; Kaushal, 2008; Potochnick, 2014) and building upon some elements of those studies focused on the estimation of differentiated state effects (Flores, 2010a; Flores & Chapa, 2009), the aim is to isolate the independent effects of policies prohibiting the access to ISRT for UIS on three educational outcomes: the college participation rates of UIS, the high school dropout rates among UIS, and the college participation rates of native and naturalized citizens.

The Data

An advantage of the present analysis is the principal source of information employed. Unlike the majority of previous works, we employ the ACS rather than the CPS.⁵ According to Passel and Cohn (2014) and B. C. Baker and Rytina (2013), the larger sample size of this survey produces much more precise estimates of unauthorized immigrants at the state level. It also permits a more detailed analysis of the conditions and characteristics of unauthorized immigrants. In addition to the ACS, complementary databases on state unemployment rates and minimum wages were obtained from the Bureau of Labor Statistics and the Wage and Hour Division at the U.S. Department of Labor, respectively (Bureau of Labor Statistics, 2014; U.S. Department of Labor, n.d.). Finally, data on in-state tuition rates and fees at postsecondary education institutions were provided by the College Board.

The ACS provides two key variables with which to approximate the population of main interest in this research, unauthorized immigrants. First, the survey classifies all individuals according to the place of birth—native born and foreign born. Second, foreign-born respondents are sorted into three groups—born abroad of U.S. parents, naturalized citizen, or not a citizen. The intersection of two characteristics, foreign born and noncitizen, is the basis for construction of the proxy variable for unauthorized immigrants. Also, the ACS collects immigration variables such as country of birth and year of immigration that contribute to the identification of the unauthorized immigrant population. The objective is to include in the analysis individuals who are highly likely unauthorized immigrants. This group consists of FBNCs who (a) entered the United States after 1981, (b) were self-identified as Hispanics, (c) were 15 years old or younger at the time of entrance, and (d) were age 16 to 24 when they were interviewed (Ruggles et al., 2010).⁶ Nevertheless, a limitation of using a proxy to identify the population of main interest, that is, unauthorized immigrants, is that the resulting policy effects are likely to be underestimated because authorized immigrants (as those present with a valid visa and green card holders) not affected by the policy are included in the ACS sample of Hispanic FBNC. Also, the undercount of unauthorized immigrants in the ACS makes the estimated effects of banning access to ISRT for UIS downward biased. The U.S. Department of Homeland Security assumed an undercount rate for unauthorized immigrants in the ACS of 10% (B. C. Baker & Rytina, 2013). Estimates of unauthorized immigrant population by B. C. Baker and Rytina (2013) indicate that ACS samples of FBNCs would be contaminated approximately 14% due to the inclusion of nonimmigrant population with legal permits (students, temporary workers, and exchange visitors). Similar to Kaushal (2008) calculations, these figures would yield a rough 22.6% downward bias in the estimates of policy effects on Hispanic FBNC highly likely unauthorized immigrants—14% due to the presence of legal immigrants and 8.6% due to the undercount of unauthorized immigrants.

The analysis covers years 2005 through 2012 because of the availability of information. The ACS was fully implemented in 2005, representing 1% of the total population. Previous smaller sample issues are not comparable with data released since 2005. As of this writing, the last year of Public Use Microdata Samples (PUMS) published by the Census Bureau is 2012.

Method

To answer the research questions, a triple-differences identification strategy is advanced through the construction of a natural quasi-experiment (Cameron & Trivedi, 2005; Meyer, 1995). The research capitalizes on two facts: first, the enactment during years 2005 to 2012 of policies banning ISRT for UIS in five states: Arizona, Colorado, Georgia, Indiana, and Ohio; and second, the nonexistence of these types of state policies (i.e., policies neither allowing ISRT for UIS, nor banning the enrollment of this population in public colleges) in 29 states and the District of Columbia during the same period. The research exploits this state–time exogenous variability in the implementation of policies banning access to ISRT for UIS to estimate policy effects on the three outcomes of interest.

Because the implementation of ISRT policies has resulted mainly from political, economic, fiscal, and cultural factors, rather than from the response to changes in the state’s higher education outcomes for UIS or U.S.-born individuals (Dougherty, Nienhusser, & Vega, 2010; McLendon, Mokher, & Flores, 2011; Reich & Mendoza, 2008; Sanders, 2010; Thangasamy, 2007; Vargas, 2011), state laws prohibiting ISRT access for unauthorized immigrants are considered as an “exogenous source of variation in the explanatory variables that determine the treatment assignment” (Meyer, 1995, p. 151). Therefore, this policy intervention permits the configuration of a natural quasi-experiment. The sample of FBNC Hispanics living in the states with ISRT restrictions or “policy states” (Arizona, Colorado, Georgia, Indiana, and Ohio) is used as the treatment group affected by the policy, whereas the comparison group consists of a similar population group living in the “nonpolicy states,” those states that never implemented a ban or any other state policy regulating the access to ISRT for UIS between 2005 and 2012.

An estimation of the type described above requires that the time trends of the outcome variable observed in both groups, treatment and control, would had been the same in the absence of the intervention; this condition is known as the parallel trend assumption (Abadie, 2005; Cameron & Trivedi, 2005; Li, Graham, & Majumdar, 2012). However, it is not possible to directly test this assumption because it will never be known what would have happened to the treatment group in a universe without the policy intervention. Nor can it be known what would have happened to the comparison group if the policy had been implemented in their states. Fortunately, there are indirect alternatives to evaluate the validity of the parallel assumption. Chin and Juhn (2011) proposed using a sample of Hispanic legal residents, a group which, in principle, is not affected by the policy, to obtain an estimate of “the difference in outcome that would exist between [policy states] and [nonpolicy states] even if there were no such laws at all” (p. 72, emphasis in original; i.e., an estimate of the differential trend between treatment and comparison groups). In this way, if the differential found is equal to zero, the parallel assumption would be valid. In the opposite case with a nonzero differential, the estimated differential can still be used to adjust the effect of the policy on the group of affected individuals (i.e., unauthorized immigrants).

In their study of the effects of state laws granting ISRT for undocumented students, Chin and Juhn (2011) exploited the mixed citizenship status of immigrant families in which unauthorized immigrant parents have children who are either unauthorized immigrants, U.S.-born citizens, or both (Fix & Zimmermann, 2001; Passel, Lopez, Cohn, & Rohal, 2014; Taylor, Lopez, Passel, & Motel, 2011). Passel et al. (2014) estimated that in 2012, there were 775,000 unauthorized children younger than 18 whereas the number of U.S.-born children at those ages living with at least one unauthorized parent reached 4.4 million. The intuition of the strategy is that both groups, Hispanic unauthorized immigrants and U.S.-born Hispanic, share similar backgrounds and those living in a particular state also face in common the state’s economic, social, cultural, and political conditions. Because the former group is affected by the policy whereas the latter is not, U.S.-born Hispanics are an adequate comparison group to estimate the differential trend between policy and nonpolicy states for the group of unauthorized immigrants.⁷ Nevertheless, the hostile climate created by policies banning ISRT for UIS might also harm the group of Hispanic natives.

The Model

Adopting a similar strategy to the one proposed by Chin and Juhn (2011), a pooled sample of Hispanics, both likely unauthorized immigrants and U.S.-born citizens, is used to estimate the policy effect on the former group adjusted for trend differentials obtained from the latter group. The sample is also limited to individuals living in any of the policy and nonpolicy states to estimate the following triple-differences model:

\begin{array}{l} Y_{i s t j} = α_{j} + δ B A N_S T A T E_{s t} \times I (j = 1) + \\ γ B A N_S T A T E_{s t} + θ_{s j} + \\ λ_{t j} + X_{i s t j}^{'} ρ_{j} + ε_{i s t j} . \end{array}

The outcome variable $Y_{i s t j}$ varies in four dimensions: individual (i), state (s), year (t), and group affected or unaffected by the policy (j). It is a binary variable defined accordingly to each of the three research questions.⁸ Therefore, $Y_{i s t j}$ is equal to 1 if individual i living in state s presents the condition defined by the research question at hand in year t, 0 otherwise. The $B A N_S T A T E_{s t}$ is a binary variable equal to 1 if state s bans the access to ISRT for UIS at year t, 0 otherwise.⁹ $I (\cdot)$ is the indicator variable equal to 1 if an individual is in the category of foreign-born not-a-citizen defined as the group affected by the banning ISRT policy and consisting of individuals highly likely to be unauthorized immigrants, 0 otherwise.¹⁰ $B A N_S T A T E_{s t} \times I (j = 1)$ is the interaction term indicating the group targeted by the policy intervention living in those states that implemented the ISRT restriction after they effectively implemented it, 0 otherwise. The parameter associated with this term ( $δ$ ) is the main coefficient of interest in this study. It measures the effect of the state laws on the education outcomes of unauthorized immigrants adjusted for the differential trend using the group of Hispanic U.S.-born citizens. Individual- and state-level control variables that may affect the outcome variables are included ( $X_{i s t j}^{'}$ ) as well as state-specific affected group effects ( $θ_{s j}$ ) and time-varying affected group effects ( $λ_{t j}$ ) that, respectively, control for differences between states that remain over time and differences across time common to all states that are allowed to vary by j (Angrist & Pischke, 2009; Chin & Juhn, 2011). $ε_{i s t j}$ represents the random error term.

Even though we control for immigration variation, cross-state mobility of unauthorized immigrants may be a potential threat to our identification. For instance, UIS could move from states with restrictive policies on access to ISRT, to states where they have access to this educational benefit making our estimation of policy effects downward biased. However, previous research indicates that states subsidies are not a determinant factor of new immigrants’ location (Kaushal, 2005; Zavodny, 1999).

To assure the independence of treatment in the control group (Li et al., 2012), the states included in the comparison group are required to have not implemented any type of policy either granting or banning access to in-state tuition for undocumented students from 2005 to 2012. Consequently, Alabama and South Carolina are excluded from the analysis. In 2011 and 2008, respectively, they implemented policies banning the enrollment of UIS in public postsecondary institutions. Also, the 15 states that at some point during the period of analysis apply policies granting ISRT for UIS are excluded (see Table A1).¹¹

This method controls for an individual’s sociodemographic characteristics and for state-level factors that can affect educational outcomes; the former can contribute to the estimate’s precision whereas the latter helps to reduce omitted variable bias (Angrist & Pischke, 2009).¹²

In addition, to control for state educational trends that can bias the effect of the policy (for instance, the rising trend in the number of Pell Grant recipients that took place during the period of analysis [College Board, 2016]), we include the percentage of non-Hispanic Whites ages 30 to 54 with at least some college experience. The percentage of Hispanics (aged 30–54) with a high school diploma is also included to control for educational aspirations of this group (Kaushal, 2008; Potochnick, 2014).

Findings

Descriptive Analysis

The first approximation that is advanced to analyze the effects of banning access to ISRT rates for UIS is to perform t tests to evaluate whether the education outcomes, individuals’ characteristics, and state conditions differ before and after the policy is implemented in both policy and nonpolicy states, distinguishing between the group of Hispanic FBNC and Hispanic U.S.-born citizens.¹³ For the group of nonpolicy states, years 2005 through 2008 are used as prepolicy and 2009 through 2012 as postpolicy. The former period coincides with the adoption of bans in two states whereas the remaining three policy states adopted bans during the latter period. Table 1 presents the results for variables at the individual level. In comparing the education outcomes before and after state laws banning access to ISRT for UIS were implemented, the results show that the treatment group, Hispanic FBNC living in the policy states (columns 1 and 2), experienced a reduction of 3.8 percentage points in the college enrollment rate, whereas the control group living in nonpolicy states (columns 5 and 6) exhibited an increase of 1.14 percentage points. These figures provide support for the hypothesis that the policies affect negatively the chances of attending institutions of higher education for UIS; the differences are not statistically significant though. However, the group of Hispanic U.S.-born citizens (columns 3 and 4) shows a pre–post 0.7 percentage points nonsignificant increase in the policy states whereas in the nonpolicy states (columns 7 and 8) a positive significant change of 2.44 percentage points for this group.

Table 1

Summary Statistics: Educational Outcomes and Individual Characteristics by Policy and Nonpolicy States

	Policy states				Nonpolicy states
	Hispanic FBNC		Hispanic U.S.-born citizens		Hispanic FBNC		Hispanic U.S.-born citizens
	Prepolicy	Postpolicy	Prepolicy	Postpolicy	Prepolicy	Postpolicy	Prepolicy	Postpolicy
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
Educational outcomes
Attending college	0.2699 (0.0200)	0.2319 (0.0131)	0.4144 (0.0105)	0.4214 (0.0069)	0.3240 (0.0106)	0.3354 (0.0102)	0.5001 (0.0065)	0.5245*** (0.0056)
Drop out of school^a	0.2450 (0.0157)	0.1890*** (0.0116)	0.1007 (0.0061)	0.0907 (0.0038)	0.1884 (0.0083)	0.1465*** (0.0078)	0.0862 (0.0038)	0.0571*** (0.0027)
Individual characteristics
Age	21.0060 (0.0821)	21.0206 (0.0574)	20.9588 (0.0379)	20.8992 (0.0250)	20.8318 (0.0428)	20.9493** (0.0376)	20.8233 (0.0233)	20.7816 (0.0198)
Female	0.4853 (0.0220)	0.5072 (0.0150)	0.5137 (0.0101)	0.4976 (0.0066)	0.4576 (0.0110)	0.4461 (0.0105)	0.5034 (0.0063)	0.5042 (0.0053)
Married	0.2410 (0.0189)	0.2193 (0.0129)	0.1468 (0.0075)	0.1338 (0.0049)	0.1906 (0.0095)	0.1551*** (0.0077)	0.1410 (0.0045)	0.1055*** (0.0035)
Employed	0.6613 (0.0196)	0.5706*** (0.0150)	0.6589 (0.0099)	0.6352** (0.0066)	0.6594 (0.0110)	0.6163*** (0.0104)	0.6521 (0.0061)	0.6008*** (0.0053)
Speaks English^b
Does not speak English	0.0500 (0.0101)	0.0373 (0.0056)	0.0034 (0.0011)	0.0017 (0.0005)	0.0585 (0.0057)	0.0398** (0.0044)	0.0029 (0.0007)	0.0007*** (0.0002)
Yes, but not well	0.1470 (0.0161)	0.0718*** (0.0079)	0.0176 (0.0026)	0.0098*** (0.0013)	0.0870 (0.0065)	0.0861 (0.0061)	0.0106 (0.0013)	0.0090 (0.0010)
Yes, speaks very well	0.5552 (0.0231)	0.6915*** (0.0145)	0.3549 (0.0105)	0.4182*** (0.0073)	0.6518 (0.0115)	0.6492 (0.0107)	0.5272 (0.0067)	0.5327 (0.0058)
Living in an MSA	0.8140 (0.0179)	0.8358 (0.0123)	0.8226 (0.0082)	0.8441** (0.0055)	0.8984 (0.0076)	0.8979 (0.0071)	0.8367 (0.0052)	0.8411 (0.0043)
Individual level N	908	1,600	3,894	9,097	2,957	3,454	9,827	14,613
N for dropouts^a	1,303	1,869	4,373	9,932	4,030	3,808	10,999	14,748

Source: American Community Survey 2005 to 2012, 1-year samples.

Note. The sample consists of Hispanics (not including Puerto Ricans) aged 18 to 24 with high school diploma but not a bachelor’s degree living in policy and nonpolicy states in years 2005 to 2012. FBNC = foreign-born noncitizen; MSA = metropolitan statistical area.

Dropout of high school is calculated using a sample of Hispanic (not including Puerto Ricans) aged 16 to 19. Standard errors in parenthesis.

Speaks English variable also includes the following two categories not shown in the table: speaks only English and speaks well.

p < .1. **p < .05. ***p < .01.

With respect to the second outcome, dropping out of school, all pre–post differences are negative and three of them are statistically significant, suggesting better performance by both groups of Hispanics. The largest changes are for Hispanic FBNC with a reduction of 5.6 and 4.19 percentage points in policy and nonpolicy states, respectively. Hispanic citizens showed reductions of 1.0 and 2.91 percentage points. These results are consistent with national trends showing a continuous reduction in the dropout rates of Hispanics, falling from 30% at the end of the 1990s to 13% in 2012 (U.S. Department of Education, National Center for Educational Statistics, 2014). The pre–post year differences in this outcome indicate a greater improvement of Hispanic FBNC in policy states, which is inconsistent with the second hypothesis (i.e., that making UIS ineligible for ISRT may discourage them from finishing secondary education); however, as can be observed in the panel of individual characteristics in Table 1, there are differences among the groups and, over time, that must be incorporated for a more reliable estimate of the effects of the policy. Also, differences in state conditions over time and between the two groups of states must also be incorporated into the analysis (as results in Table A2 indicate).¹⁴

Multivariate Analysis

Policy Effects on the College Participation Rates of UIS

To estimate the causal effect of banning ISRT rates for UIS on their college participation rates, a linear probability model (ordinary least squares [OLS]) is used (Angrist & Pischke, 2009). Table 2 presents the estimated coefficients and robust standard errors of five specifications of the model in Equation 1 using a sample of Hispanics aged 18 through 24 with educational attainment equal to or more than a high school diploma, but less than a bachelor’s degree, living in policy and nonpolicy states. The dependent variable is equal to 1 if an individual is currently attending college, 0 otherwise. The baseline model in column (1) does not include individual or state control variables ( $X_{i s t j}^{'}$ ). Models in columns (2) and (3) add subsequently individual and state characteristics. Robust standard errors clustered by state-year and treatment group-state-year are included in column (4) and column (5), respectively. Panel A shows the policy effects on college attendance of Hispanic FBNC highly likely unauthorized immigrants adjusted for the differential trends using the estimated policy effects on U.S.-born Hispanics shown in Panel C. Finally, Panel B presents unadjusted estimated policy effects on Hispanic FBNC highly likely unauthorized immigrants. Therefore, Panels B and C show simpler difference-in-differences estimates of policy effects among noncitizens and citizens, respectively, comparing policy and nonpolicy states over time.¹⁵

Table 2

Estimated Effects of State Laws Banning Access to ISRT Rates for Unauthorized Immigrants on College Attendance

Variables	(1)	(2)	(3)	(4)	(5)
Variables	Basic model	Individual characteristics	State characteristics	Clustered by state-year	Clustered by treatment group-state-year
A. Hispanic FBNC highly likely unauthorized immigrants (effects adjusted for differential trends)
Policy effect	−0.0593* (0.0326)	−0.108*** (0.0306)	−0.0840** (0.0341)	−0.0840*** (0.0286)	−0.0840** (0.0342)
Observations	46,350	46,350	46,209	46,209	46,209
B. Hispanic FBNC highly likely unauthorized immigrants
Policy effect	−0.0329 (0.0284)	−0.0731*** (0.0266)	−0.0490* (0.0297)	−0.0490 (0.0304)	N/A
Observations	8,919	8,919	8,898	8,898	N/A
C. Hispanic U.S-Born
Policy effect	0.0264 (0.0161)	0.0349** (0.0151)	0.0349** (0.0169)	0.0349** (0.0158)	N/A
Observations	37,431	37,431	37,311	37,311	N/A

Source: American Community Survey 2005 to 2012, 1-year samples.

Note. Robust SE in parentheses. Data weighted using “perwt” IPUMS weights. Each subgroup of estimates is from a separate linear regression. All models in Panel B and C include state- and year-fixed effects. Models (2) to (5) control for age, gender, marital status, employment status, and English proficiency. Models (3) to (5) additionally control for metropolitan area, state unemployment rate, state real minimum wage, average in-state tuition and fees in 2- and 4-year colleges, proportion of non-Hispanic White adults with at least some college, and proportion of Hispanic with high school diploma. Models (3) to (5) exclude the District of Columbia because the lack of information on college tuition and fees. ISRT = in-state resident tuition; FBNC = foreign-born noncitizen; IPUMS = Integrated Public Use Microdata Series.

p < .1. **p < .05. ***p < .01.

The preferred model, column (5) in Panel A, shows that policies banning in-state resident tuition to UIS are associated with a statistically significant 8.4 percentage points decrease in the college enrollment among Hispanic FBNCs who are highly likely unauthorized immigrants (p < .05). This effect represents a 31.1% reduction in enrollment over the base-level enrollment of 26.99% reported in Table 1. Models (2) to (5) in Panel A consistently report statistically significant negative policy effects, however, including state characteristics in Model (3) reduces the estimated coefficient, and clustering by state-year (Model (4)) as well as treatment group-state-year (Model (5)) affects the levels of statistical significance. Model (4) in Panel B shows no significant policy effects on the enrollment of Hispanic FBNC highly likely unauthorized immigrants, whereas Panel C, using the group of Hispanic U.S.-born citizens, indicates that the assumption of parallel trends is not met, that is, there would exist statistically significant differences in college enrollment between policy and nonpolicy states even if policies banning ISRT for UIS there were not implemented at all. The positive sign of the effects in Panel C means that without policies banning ISRT, college attendance among Hispanics would have increased in policy states relative to nonpolicy implying that adjusted estimates in Panel A show larger negative effects than those in Panel B.

Because they are the group with highest probability of being unauthorized, much of previous research on the educational effects of ISRT policies has focused on unauthorized immigrants coming from Mexico. For consistency with previous research, the final model is fit using the complete sample of Hispanics as well as a subsample consisting only of Mexicans with both samples including FBNCs and U.S. citizens. Table 3 presents the results by sex and by age. Panel A with pooled samples of men and women shows that the effect of the policy is comparable for all Hispanic FBNC and for the subsample of Mexican FBNCs. Therefore, the college enrollment among the subgroup of younger individuals (ages 18–20) of Hispanic and Mexican FBNC in policy states is, respectively, 12.7 (p < .05) and 12.4 (p < .05) percentage points lower after laws banning ISRT took effect, relative to the comparison groups in nonpolicy states. Differentiating by sex, Panels B and C indicate that the largest effect of the policy is on the enrollment of younger Mexican men, 17.3 percentage points reduction (p < .01), and marginal significant effects are found among the subgroup of Hispanic women aged 21 through 24 (9.5 percentage point reduction; p < .1).

Table 3

Estimated Effects of State Laws Banning Access to ISRT Rates for Unauthorized Immigrants on College Attendance of Hispanic and Mexican FBNC Highly Likely Unauthorized Immigrants Adjusted for Differential Trends Using Hispanic and Mexican U.S.-Born Citizens

	Hispanics			Mexicans
	Policy effect	(SE)	Observations	Policy effect	(SE)	Observations
A. Men and women
Ages 18–24	−0.0840**	(0.0342)	46,209	−0.0538	(0.0357)	25,476
Ages 18–20	−0.127**	(0.0526)	21,974	−0.124**	(0.0480)	11,785
Ages 21–24	−0.0674*	(0.0384)	24,235	−0.0169	(0.0443)	13,691
B. Only men
Ages 18–24	−0.0829*	(0.0465)	23,195	−0.0808*	(0.0468)	12,893
Ages 18–20	−0.156**	(0.0711)	10,800	−0.173***	(0.0617)	5,904
Ages 21–24	−0.0490	(0.0492)	12,395	−0.0230	(0.0550)	6,989
C. Only women
Ages 18–24	−0.0818*	(0.0420)	23,014	−0.0235	(0.0491)	12,583
Ages 18–20	−0.0885	(0.0728)	11,174	−0.0780	(0.0749)	5,881
Ages 21–24	−0.0952*	(0.0568)	11,840	−0.0100	(0.0666)	6,702

Source: American Community Survey 2005 to 2012, 1-year samples.

Note. Robust SE in parenthesis from clustering by treatment group-state-year. Data weighted. Each subgroup of estimates is from a separate linear regression that controls for age, gender, marital status, employment status, English proficiency, metropolitan area, state unemployment rate, state real minimum wage, average in-state tuition and fees in 2- and 4-year colleges, proportion of non-Hispanic White adults with at least some college, and proportion of Hispanic with high school diploma. ISRT = in-state resident tuition; FBNC = foreign-born noncitizen.

p < .1. **p < .05. ***p < .01.

Policy Effects on Dropping Out of School

State policies banning access to ISRT for UIS may discourage unauthorized immigrant youth from finishing high school because for many of them, the possibility of advancing to the next level of education and having to pay out-of-state tuition is slim, and thus, would make obtaining a high school diploma less valuable. Estimates of the policy effects on dropping out for both all Hispanic FBNCs and Mexican FBNCs aged 16 through 19 using the model represented in Equation 1 are presented in Table A3. The estimates indicate that after ISRT bans were implemented, Hispanic FBNCs and Mexican FBNCs in the sample are less likely to drop out of school; however, the results are not statistically significant. Once the samples are divided in two subgroups, aged 16 to 17 and 18 to 19, the estimated policy effects on Hispanic and Mexican FBNCs are positive for the younger group (2.4 and 4.3 percentage points) and negative for the older (−1.9 and −3 percentage points); these results are not statistically significant though.

Policy Effects on U.S. Citizens

The effects of state laws making UIS ineligible for ISRT can extend beyond the target group. Having a reduced demand for postsecondary education from UIS might mean that public institutions can serve more citizens and legal immigrants. To evaluate this possibility, a similar model as that specified in note 15 is fitted using samples of non-Hispanic individuals to estimate the policy effects on three groups of citizens—Whites, Blacks, and Asians. In addition, effects on Hispanic and Mexican naturalized citizens are evaluated. Table 4 presents the estimated policy effects. The results show that statistically significant effects are found only for the group of non-Hispanic White citizens. After ISRT policies were implemented, the college attendance among those living in policy states falls 0.97 percentage points (p < .05), compared with otherwise similar group of people living in nonpolicy states.

Table 4

Estimated Effects of State Laws Banning Access to ISRT Rates for Unauthorized Immigrants on College Attendance of U.S. Citizens

	Naturalized citizens		Non-Hispanic
	Hispanic	Mexicans	Whites	Blacks	Asians
Policy effect	−0.0775 (0.0494)	−0.0542 (0.0554)	−0.00974** (0.00396)	0.00995 (0.0114)	−0.0267 (0.0254)
Observations	4,423	1,364	533,378	90,009	12,315
R ²	.206	.205	.202	.119	.254

Source: American Community Survey 2005 to 2012, 1-year samples.

Note. Robust SE in parenthesis from clustering by state-year. Data weighted. Each subgroup of estimates is from a separate linear probability model that controls for age, gender, marital status, employment status, English proficiency, metropolitan area, state unemployment rate, state real minimum wage, state average in-state tuition and fees in 2- and 4-year colleges, proportion of non-Hispanic White adults with at least some college, and proportion of Hispanic with high school diploma; includes state- and year-fixed effects. ISRT = in-state resident tuition.

p < .1. **p < .05. ***p < .01.

Robustness Tests

To test the possibility of bias in the results presented above, three tests are advanced. First, lead and lagged placebo policies are used to estimate the final model (Model 5 from Table 2) to test the dynamics of the effects of the ISRT policy and Granger causality (i.e., that causes happen before consequences). Second, the final model is fitted including specific state linear trends. Finally, the effect of the policy is estimated on an alternative group of noncitizens (Angrist & Pischke, 2009; Flores, 2010b; Kaushal, 2008; Meyer, 1995; Potochnick, 2014).

To determine whether the ISRT policies happened before its effects take place, the $B A N_S T A T E_{s t}$ variable in Equation 1 is defined with 1-year lead. To examine the effects as time passes, the variable is defined with 1-, 2-, and 3-year lags.¹⁶ The estimated policy effects for the four placebo policies and the actual ISRT policy are plotted in Figure 1. The estimates indicate nonstatistically significant changes in college attendance among UIS living in policy states in comparison with those in nonpolicy states during the year prior to ISRT implementation. The lack of statistical significance of the estimated coefficient ( $β$ = −.0300, SE = 0.0354, p = .369, n = 46,209) indicates that the effect of the ISRT policy on Hispanic FBNC during the year of adoption is not confounded with previous state-specific trends and suggests that the ISRT policy causes the changes in attendance rates in college and not vice versa (Angrist & Pischke, 2009; Autor, 2003; Kaushal, 2008). However, the largest effect of the policies is experienced 1 year after adoption of the five state policies ( $β$ = −.1264, SE = 0.0349, p < .0001, n = 46,209). Decreasing effects in the subsequent two periods are found. However, the policy effects are statistically significant 2 years after policy adoption only ( $β$ = −.0747; SE = 0.0336, p < .05, n = 46,209). These results show the dynamics of the postpolicy effects. Therefore, college attendance among Hispanic FBNCs in the group of policy states is lower than it was before policy implementation, but these effects appear to attenuate over time. The lack of statistical significance after 3 years of policy implementation may be explained by the reduction in the number of postpolicy observations as result of lagging the policy intervention.

Figure 1.

Estimated effects before, during, and after ISRT policies adoption.

The second robustness check is to incorporate state-specific linear trends into the full model to allow policy and nonpolicy states to follow differentiated trends. Despite the fact that Equation 1 includes some variables to control for state education trends and FBNC population, it is still possible that unobserved state-specific time trends could confound the results. If adding the new controls to the model changes the policy effects obtained by the final model, it would be an indication of the presence of bias (Angrist & Pischke, 2009; Kaushal, 2008). This robustness check suggests that the policy effects found by the main model ( $β$ = −.0840, SE = 0.0342, p = .014, n = 46,209) are slightly modified by the inclusion of state-specific time trends ( $β$ = −.0789, SE = .0233, p = .001, n = 46,209).

Finally, to build confidence that the findings apply to unauthorized immigrants and not to other groups of immigrants, the main model is fitted using a sample of Asian U.S. citizens and Asian FBNCs. The latter group fulfills the conditions of age (18–24), year of entrance to the United States (1982 or later), age at entrance to the United States (15 or younger), and educational attainment (high school diploma, but not bachelor’s degree) applied to the sample used to estimate the policy effects on Hispanic FBNCs (highly likely unauthorized immigrants). If it is found that policy effects using the sample of Asians are statistically different from zero, it would indicate that banning ISRT policies can be correlated with other unobservable factors that affect all FBNC and not exclusively highly likely unauthorized immigrants. Using a similar model specification as in column (5) of Table 2, the results indicate that Asian FBNC youth are not affected by the ISRT policy ( $β$ = .1445, SE = 0.1116, p = .196), which provides more confidence in the findings for Hispanic and Mexican FBNCs highly likely to be unauthorized in the United States.

Summary and Discussion

The policies analyzed in this research were implemented in Arizona and Colorado in 2006, Georgia in 2008, and Indiana and Ohio in 2011. None of the policies completely closes the door to higher education for unauthorized immigrants. This population can enroll in public institutions in these states as long as they pay out-of-state tuition rates. However, the trinity of difficult economic conditions (Fortuny, Capps, & Passel, 2007; López, 2010; Passel & Cohn, 2009), high costs of education (Abrego & Gonzales, 2010; López, 2010), and immigration status that makes them ineligible for federally funded aid and in some states for education benefits funded with public resources (Biswas, 2005; Ruge & Iza, 2005; Salinas, 2006), prevent most unauthorized immigrants from enrolling in college. How do state policies that explicitly ban the access to ISRT for UIS affect college attendance among this population? This research found evidence (p < .05) that supports the first research hypothesis. Therefore, after state laws making UIS ineligible to pay ISRT rates were implemented, college attendance among Hispanic FBNCs highly likely unauthorized immigrants aged 18 to 24 fell 8.4 percentage points during the implementation year, compared with the control group of people living in states without these policies. The largest effect of the policy, 12.6 percentage point decrease, takes place the year after adoption of the measure. The results also indicate that the policies mainly harm younger students (18–20-year-olds) whose college enrollment decreases 12.7 percentage points among Hispanic UIS. Restricting the focus to Mexican FBNCs, the group most affected by the ISRT prohibition consists of men 18 to 20 years old. This group experienced a 17.3 percentage point decrease in college attendance as result of the policies. However, no evidence is found for Mexican women in any of the age ranges. Nonetheless, the direction of the effects on all subgroups of FBNC highly likely unauthorized immigrants is consistent with our expectations.

To compare our results with those of previous studies on the impact of ISRT policies, it is important to emphasize that prior studies were looking at the effect of policy introduction whereas our research looks at the effect of policy elimination. Therefore, we would expect the impacts of the two types of policies to be opposite in sign. Kaushal (2008) and Amuedo-Dorantes and Sparber (2014) estimated, respectively, a 2.5 percentage points (31%) and a 4 percentage point (23%) increase in college enrollment among Mexican FBNCs. Consistently, we find that Mexican FBNCs aged 18 to 20 experienced a 12.4 percentage points (42.9%) reduction in college enrollment as result of state policies banning ISRT to them, however, nonsignificant effects were found among the group of older Mexican FBNCs. With regard to the effects of policies on a broader group of people, FBNC Hispanics, Dickson and Pender (2013) found opposite results in two Texas universities (15 and −2 percentage points). Their first result is consistent with our estimated effect on this group: 8.4 percentage points (31%) fall in college attendance. Finally, the only work studying the effects of banning ISRT to UIS, similar to the policy analyzed by us though temporal, is Conger and Turner (2015). Our estimates on the group of Hispanic FBNCs show higher negative impacts, 8.4 percentage (31%), than theirs, 7 percentage points (8%). However, it is important to note that their impacts are on the reenrollment of New York City’s public university system senior UIS, whereas our estimates are on college enrollment of Hispanic FBNCs in five policy states.

Why are the effects of the policies larger for the group of younger individuals? Having restricted the population of interest in this research—the highly likely unauthorized immigrants—to those Hispanic FBNCs with a high school diploma who came to the United States at age 15 or younger, the results on the younger group indicate that policies banning ISRT have affected mainly U.S. high schools’ recent graduates. Chin and Juhn (2011) offer insights to explain these differences by age ranges. First, between the two groups of students, assuming both of them have the same desire to attend college, the older group would face fewer credit constraints because they have had the opportunity to work and save for college whereas the younger group depends mainly on their parents’ resources. Therefore, human capital theory would predict that state policies making UIS ineligible for ISRT would have larger effects on younger students through higher monetary cost and larger credit constraints relative to older students. Second, the group of younger individuals may exhibit higher nonmonetary costs associated with the risk of deportation. Because recent high school graduates are more likely to live with their parents, they are more reluctant to share information with colleges and universities about their immigration status or that can link them to their families.

An alternative explanation for the dissimilar effects by age is that younger and older individuals may value present and future consumption differently. Coming from poor families, Hispanic FBNCs highly likely unauthorized immigrants of college age are usually first-generation students without a family role model to follow (Gonzales, 2010; Perez, Espinoza, Romas, Coronado, & Cortes, 2009; Yoshikawa, 2010). Furthermore, economic pressures can lead recent high school graduates to participate in the labor market rather than to enroll in higher education (i.e., present is valued more than future consumption). However, at older ages, individuals may realize the value of education and the future benefits it would bring, encouraging some to return to school. If this were the case, individuals who value future consumption more would be less affected by policies that make them to pay higher prices for education.

The second research hypothesis, which relates to high school dropout rates, is not supported. An explanation for the lack of dynamic effects may be that during school years unauthorized immigrants are not fully aware of their immigration status or of its implications in terms of access to college education. Also, it is possible that the effects of other state education policies seeking to reduce high school dropout rates are confounded with the effects of policies banning ISRT. Because the model fitted in this research has no control for the presence of other contemporaneous policies, these policies might offset the actual effect of ISRT prohibitions on unauthorized immigrants of school age. A final possibility might be that after policies banning ISRT for UIS are implemented, this group might see high school graduation as more important as well as more achievable than a college degree goal because it would be the highest available education attainment.

Supporters of state policies banning access to ISRT for UIS would find support if these measures benefit U.S.-born citizens. However, we find no positive policy effects on non-Hispanic White, Black, and Asian citizens or on Hispanic and Mexican naturalized citizens. With regard to Hispanic U.S.-born citizens, the statistically significant differential trends obtained above (Panel C in Table 2) may be interpreted as college attendance benefits associated with the ISRT policy among this group rather than as an estimated of the differential trend. The new policies may make this population more aware of and value more highly the higher education benefits already available to them, inducing them to enroll in postsecondary education.

Despite the methodological strategies implemented in this research, to isolate the causal policy effects on the three outcomes studied, two limitations remain. First, the lack of control for other contemporaneous state policies that could affect unauthorized immigrants (for instance, drivers’ licenses, access to health benefits, and law enforcement measures) may indirectly affect education outcomes, and consequently, the effects identified here can be confounded with other policies. Second, the period of analysis determined by the availability of data from the main source of information, the ACS, limits the research to only two prepolicy years of information for the first two states with ISRT prohibitions (Arizona and Colorado), and only one postpolicy year of data for the last two states implementing the policy (Indiana and Ohio). The policy effects may vary depending on how long the policies have been in place, as the placebo policy analysis showed.

Comparing the effects of states’ legislative actions prohibiting UIS access to ISRT, on the Hispanic FBNC (who are highly likely to be unauthorized immigrants) and U.S.-born citizens, the research indicates that the effects were negative for the former group with no gains for the latter group. Because federal laws (IIRIRA, 1996; PRWORA, 1996) already made unauthorized immigrants ineligible for in-state tuition save for the cases in which the states themselves advance laws to provide that benefit, the bans implemented in Arizona, Colorado, Georgia, Indiana, and Ohio seem to have no explicit legal purpose. Our results show important negative effects on college attendance for this population. In addition, the cost of the banning ISRT policies has been born mainly by recent high school graduates aged 18 to 20. The expenditure per student enrolled in public elementary and secondary schools was US$12,401 in 2011–2012 school year (U.S. Department of Education, National Center for Educational Statistics, 2015). ISRT bans are inefficient because the policies truncate the education aspirations of some of those who previously were served by the U.S. public education system.

This research informs policy and practice, not only for those states that have implemented the ISRT restrictions, but also for states that are considering the adoption of such measures. Unauthorized immigrants exhibit the poorest education outputs (college attendance and school dropout rates) studied here. Unauthorized immigrants do even worse in the group of policy states after ISRT prohibitions were implemented. In an era characterized by rapid technological changes, increasing demand for skilled labor, and global competition, states would benefit from having a more educated population, including unauthorized immigrants who might not relocate. Facilitating access to higher education by reevaluating and revoking previous state decisions such as those taken by the five policy states examined in this research would be a first step toward more efficient, fairer, and more productive state education systems. This research provides new insights on the debate about issues of access to ISRT for UIS that will hopefully contribute to the future development of state policies.

Footnotes

Appendix

Table A3

Estimated Effects of State Laws Banning Access to ISRT Rates for Unauthorized Immigrants on the Dropping Out From School of Hispanic and Mexicans FBNC Highly Likely Unauthorized Immigrants Adjusted for Differential Trends Using Hispanic and Mexican U.S.-Born Citizens

	(1)	(2)
	Hispanics	Mexicans
A. Ages 16–19
Policy effect	−0.00527 (0.0185)	−0.00275 (0.0249)
Observations	50,933	29,395
B. Ages 16–17
Policy effect	0.0240 (0.0204)	0.0427* (0.0251)
Observations	26,856	15,678
C. Ages 18–19
Policy effect	−0.0186 (0.0350)	−0.0305 (0.0477)
Observations	24,077	13,717

Source: American Community Survey 2005 to 2012, 1-year samples.

Note. Robust SE in parentheses from clustering by treatment group-state-year. Data weighted using “perwt” IPUMS weights. Each subgroup of estimates is from a separate linear regression and controls for age, gender, marital status, employment status, and English proficiency, metropolitan area, state unemployment rate, state real minimum wage, average in-state tuition and fees in 2- and 4-year colleges, proportion of non-Hispanic White adults with at least some college, and proportion of Hispanic with high school diploma. Estimations exclude the District of Columbia because of the lack of information on college tuition and fees.

p < .1. **p < .05. ***p < .01.

Acknowledgements

Alexander Villarraga-Orjuela is grateful to Kathleen A. Lehman for her valuable help. The authors would like to thank Aimee Chin and two anonymous referees.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

Authors

ALEXANDER VILLARRAGA-ORJUELA is professor of economics and director of the Observatorio de Educación at the Universidad del Norte, Colombia. His research examines policies relating equity and quality in education mainly on the access, success, and completion in higher education among disadvantaged groups. Contact information: Economics Department, Universidad del Norte, Colombia, Bloque D 215, Barranquilla, Colombia (avillarraga@uninorte.edu.co).

BRINCK KERR is professor of political science and director of the Public Policy PhD Program at the University of Arkansas. His research focuses on representative bureaucracy, race and ethnicity, education policy, and employment policy. Contact information: Department of Political Science, University of Arkansas, Old Main 428, Fayetteville, AR 72701 (jbkerr@uark.edu).

References

Abadie

(2005). Semiparametric difference-in-differences estimators. The Review of Economic Studies, 72, 1–19. doi:10.2307/3700681

Abrego

L. J.

(2006). “I can’t go to college because I don’t have papers”: Incorporation patterns of Latino undocumented youth. Latino Studies, 4, 212–231.

Abrego

L. J.

Gonzales

R. G.

(2010). Blocked paths, uncertain futures: The postsecondary education and labor market prospects of undocumented Latino youth. Journal of Education for Students Placed at Risk, 15, 144–157.

Amuedo-Dorantes

Sparber

(2014). In-state tuition for undocumented immigrants and its impact on college enrollment, tuition costs, student financial aid, and indebtedness. Regional Science & Urban Economics, 49, 11–24. doi:10.1016/j.regsciurbeco.2014.07.010

Angrist

J. D.

Pischke

J.-S.

(2009). Mostly harmless econometrics: An empiricist’s companion. Princeton, NJ: Princeton University Press.

Autor

D. H.

(2003). Outsourcing at will: The contribution of unjust dismissal doctrine to the growth of employment outsourcing. Journal of Labor Economics, 21, 1–42. doi:10.1086/344122

Baker

B. C.

Rytina

(2013). Estimates of the unauthorized immigrant population residing in the United States: January 2012. Retrieved from http://www.dhs.gov/sites/default/files/publications/ois_ill_pe_2012_2.pdf

Baker

(1997). The “Amnesty” aftermath: Current policy issues stemming from the legalization programs of the 1986 Immigration Reform and Control Act. International Migration Review, 31, 5–27. doi:10.2307/2547255

Batalova

Mittelstadt

(2012). Relief from deportation: Demographic profile of the DREAMers potentially eligible under the Deferred Action Policy. Washington, DC: Migration Policy Institute.

10.

Becker

G. S.

(1962). Investment in human capital: A theoretical analysis. Journal of Political Economy, 70(5), 9–49.

11.

Becker

G. S.

(1993). Human capital: A theoretical and empirical analysis, with special reference to education (3rd ed.). Chicago, IL: The University of Chicago Press.

12.

Bell Policy Center. (2005, April). Undocumented immigrant students and access to higher education: An overview of federal and state policy (Issue brief). Denver, CO: Author.

13.

Biswas

(2005). Access to community college for undocumented immigrants: A guide for state policymakers. Boston, MA: Jobs for the Future.

14.

Bureau of Labor Statistics. (2014). Local area unemployment statistics. Retrieved from http://www.bls.gov/lau/lauov.htm

15.

Cameron

A. C.

Trivedi

P. K.

(2005). Microeconometrics: Methods and applications. Cambridge, UK: Cambridge University Press.

16.

Chin

Juhn

(2011). Does reducing college costs improve educational outcomes for undocumented immigrants? Evidence from state laws permitting undocumented immigrants to pay in-state tuition at state colleges and universities. In Leal

D. L.

Trejo

S. J.

(Eds.), Immigrants and Minorities, Politics and Policy Series: Latinos and the economy: Integration and impact in schools, labor markets, and beyond (pp. 63–94). New York, NY: Springer.

17.

Cojoc

A. E.

(2010). Essays in public economics (Unpublished 3419262). Clemson University, Clemson, SC.

18.

College Board. (2016). Total Pell Grant expenditures and number of recipients over time. Retrieved from https://trends.collegeboard.org/student-aid/figures-tables/pell-grants-total-expenditures-maximum-and-average-grant-and-number-recipients-over-time#Key%20Points

19.

Conger

Turner

L. J.

(2015, April). The impact of tuition increases on undocumented college students’ attainment (National Bureau of Economic Research Working Paper Series, No.21135). National Bureau of Economic Research. doi:10.3386/w21135

20.

Darolia

Potochnick

(2015). Educational “when,” “where,” and “how” implications of in-state resident tuition policies for Latino undocumented immigrants. Review of Higher Education, 38, 507–535. doi:10.1353/rhe.2015.0028

21.

DesJardins

Toutkoushian

(2005). Are students really rational? The development of rational thought and its application to student choice. In Smart

(Ed.), Higher education: Handbook of theory and research (Vol. 20, pp. 191–240). Dordrecht, the Netherlands: Springer Netherlands.

22.

Dickson

Pender

(2013). Do in-state tuition benefits affect the enrollment of non-citizens? Evidence from universities in Texas. Economics of Education Review, 37, 126–137. doi:10.1016/j.econedurev.2013.08.006

23.

Dougherty

K. J.

Nienhusser

H. K.

Vega

B. E.

(2010). Undocumented immigrants and state higher education policy: The politics of in-state tuition eligibility in Texas and Arizona. Review of Higher Education, 34, 123–173.

24.

Education Commission of the States. (2014). State policy database. Retrieved from https://b5.caspio.com/dp.asp?AppKey=b7f930003204329dff2e4cf1af9e&yr=2014

25.

Fix

Zimmermann

(2001). All under one roof: Mixed-status families in an era of reform. The International Migration Review, 35, 397–419.

26.

Flores

S. M.

(2010a). The First State Dream Act: In-state resident tuition and immigration in Texas. Educational Evaluation and Policy Analysis, 32, 435–455.

27.

Flores

S. M.

(2010b). State Dream Acts: The effect of in-state resident tuition policies and undocumented Latino students. Review of Higher Education, 33, 239–283.

28.

Flores

S. M.

Chapa

(2009). Latino immigrant access to higher education in a bipolar context of reception. Journal of Hispanic Higher Education, 8, 90–109.

29.

Fortuny

Capps

Passel

J. S.

(2007). The characteristics of unauthorized immigrants in California, Los Angeles County, and the United States. Washington, DC: Urban Institute.

30.

Frum

(2007). Postsecondary educational access for undocumented students: Opportunities and constraints. American Academic, 3(1), 81–108.

31.

Gonzales

(2010). On the wrong side of the tracks: Understanding the effects of school structure and social capital in the educational pursuits of undocumented immigrant students. Peabody Journal of Education, 85, 469–485.

32.

Gonzales

(2011). Learning to be illegal. American Sociological Review, 76, 602–619.

33.

Hoefer

Rytina

Baker

(2012). Estimates of the Unauthorized Immigrant Population Residing in the United States: January 2011. Retrieved from https://www.dhs.gov/xlibrary/assets/statistics/publications/ois_ill_pe_2011.pdf

34.

Illegal Immigration Reform and Immigration Responsibility Act. (1996). Illegal Immigration Reform and Immigrant Responsibility Act. In. Pub. L. No 104-208, div. C, 110 Stat. 3009-546-3009-724.

35.

Institute of Higher Education Law and Governance. (2014a). Table one: State laws allowing undocumented college students to establish residency, 2014. Retrieved from http://www.law.uh.edu/ihelg/documents/Statute-TableOne.html

36.

Institute of Higher Education Law and Governance. (2014b). Table two: States restricting access to postsecondary education, 2014. Retrieved from http://www.law.uh.edu/ihelg/documents/Statute-TableTwo.html

37.

Immigration Reform and Control Act. (1986). Immigration Reform and Control Act. In.pub. L. No. 99-603, 100 Stat.

38.

Kahneman

Tversky

(1979). Prospect theory: An analysis of decision under risk. Econometrica, 47, 263–291. doi:10.2307/1914185

39.

Kaushal

(2005). New immigrants’ location choices: Magnets without welfare. Journal of Labor Economics, 23, 59–80.

40.

Kaushal

(2008). In-state tuition for the undocumented: Education effects on Mexican young adults. Journal of Policy Analysis & Management, 27, 771–792.

41.

Graham

D. J.

Majumdar

(2012). The effects of congestion charging on road traffic casualties: A causal analysis using difference-in-difference estimation. Accident Analysis & Prevention, 49, 366–377. doi:10.1016/j.aap.2012.02.013

42.

López

(2010). Undocumented students and the policies of wasted potential. New York, NY: LFB Scholarly.

43.

McLendon

M. K.

Mokher

C. G.

Flores

S. M.

(2011). Legislative agenda setting for in-state resident tuition policies: Immigration, representation, and educational access. American Journal of Education, 117, 563–602.

44.

Meyer

B. D.

(1995). Natural and quasi-experiments in economics. Journal of Business & Economic Statistics, 13, 151–161.

45.

National Center for Educational Statistics, Digest of Education Statistics. (2013). Table 302.20. Percentage of recent high school completers enrolled in 2- and 4-year colleges, by race/ethnicity: 1960 through 2012. Retrieved from http://nces.ed.gov/programs/digest/d13/tables/dt13_302.20.asp

46.

National Conference of State Legislatures. (2011). Undocumented student tuition: Federal action. Retrieved from http://www.ncsl.org/research/education/undocumented-student-tuition-federal-action.aspx

47.

National Conference of State Legislatures. (2014a). Undocumented student tuition: Overview. Retrieved from http://www.ncsl.org/issues-research/educ/undocumented-student-tuition-overview.aspx

48.

National Conference of State Legislatures. (2014b). Undocumented student tuition: State action. Retrieved from http://www.ncsl.org/issues-research/educ/undocumented-student-tuition-state-action.aspx

49.

Olivas

M. A.

(2004). IIRIRA, the Dream Act, and undocumented college student residency. Journal of College & University Law, 30, 435–464.

50.

Olivas

M. A.

(2009). The political economy of the DREAM Act and the legislative process: A case study of comprehensive immigration reform. The Wayne Law Review, 55, 1757–1810.

51.

Passel

J. S.

(2005). Estimates of the size and characteristics of the undocumented population. Washington, DC: Pew Hispanic Center.

52.

Passel

J. S.

Cohn

(2008). Trends in unauthorized immigration: Undocumented inflow now trails legal inflow. Washington, DC: Pew Hispanic Center.

53.

Passel

J. S.

Cohn

(2009). A portrait of unauthorized immigration: Undocumented inflow now trails legal inflow. Washington, DC: Pew Hispanic Center.

54.

Passel

J. S.

Cohn

(2011). Unauthorized immigrant population: National and state trends, 2010. Washington, DC: Pew Hispanic Center.

55.

Passel

J. S.

Cohn

(2014). Unauthorized immigrant totals rise in 7 states, fall in 14. Washington, DC: Pew Hispanic Center.

56.

Passel

J. S.

Cohn

Gonzalez-Barrera

(2013). Population decline of unauthorized immigrants stalls, may have reversed. Washington, DC: Pew Research Center.

57.

Passel

J. S.

Lopez

M. H.

Cohn

Rohal

(2014). As growth stalls, unauthorized immigrant population becomes more settled. Washington, DC: Pew Research Center.

58.

Paulsen

Toutkoushian

(2008). Economic models and policy analysis in higher education: A diagrammatic exposition. In Smart

(Ed.), Higher education (Vol. 23, pp. 1–48). Dordrecht: Springer Netherlands.

59.

Perez

Espinoza

Romas

Coronado

Cortes

(2009). Academic resilience among undocumented Latino students. Hispanic Journal of Behavioral Sciences, 31, 149–181.

60.

Plyler v. Doe, 457 U.S. 202 (1982).

61.

Potochnick

(2014). How states can reduce the dropout rate for undocumented immigrant youth: The effects of in-state resident tuition policies. Social Science Research, 45, 18–32. doi:10.1016/j.ssresearch.2013.12.009

62.

Personal Responsibility and Work Opportunity Act. (1996). Personal Responsibility and Work Opportunity Act. In. Pub. L. No 104-193, 110 Stat2105.

63.

Reich

Mendoza

A. A.

(2008). “Educating kids” versus “coddling criminals”: Framing the debate over in-state tuition for undocumented students in Kansas. State Politics & Policy Quarterly, 8, 177–197.

64.

Romero

V. C.

(2002). Postsecondary school education benefits for undocumented immigrants: Promises and pitfalls. North Carolina Journal of International Law & Commercial Regulation, 27, 393–418.

65.

Ruge

T. R.

Iza

A. D.

(2005). Higher education for undocumented students: The case for open admission and in-state tuition rates for students without lawful immigration status. Indiana International & Comparative Law Review, 15, 257–278.

66.

Ruggles

Alexander

J. T.

Genadek

Goeken

Schroeder

M. B.

Sobek

(2010). Integrated Public Use Microdata Series: Version 5.0 [Machine-readable database]. Minneapolis: Minnesota Population Center.

67.

Russell

(2011). State policies regarding undocumented college students: A narrative of unresolved issues, ongoing debate and missed opportunities. Washington, DC: American Association of State Colleges and Universities.

68.

Salinas

V. J.

(2006). You can be whatever you want to be when you grow up, unless your parents brought you to this country illegally: The struggle to grant instate tuition to undocumented immigrant students. Houston Law Review, 43, 847–877.

69.

Sanders

M. S.

(2010). Hope, opportunity, and access: The in-state tuition debate in North Carolina. Journal of Latinos and Education, 9, 108–125.

70.

Shultz

T. W.

(1963). The economic value of education. New York, NY: Columbia University Press.

71.

Stevenson

(2004). Dreaming of an equal future for immigrant children: Federal and state initiatives to improve undocumented students’ access to postsecondary education. Arizona Law Review, 46, 551–580.

72.

Suárez-Orozco

Yoshikawa

Teranishi

R. T.

Suárez-Orozco

M. M.

(2011). Growing up in the shadows: The developmental implications of unauthorized status. Harvard Educational Review, 81, 438–472.

73.

Sweetland

S. R.

(1996). Human capital theory: Foundations of a field of inquiry. Review of Educational Research, 66, 341–359. doi:10.2307/1170527

74.

Taylor

Lopez

M. H.

Passel

J. S.

Motel

(2011). Unauthorized immigrants: Length of residency, patterns of parenthood. Washington, DC: Pew Hispanic Center.

75.

Thangasamy

(2007). Explaining policy making for undocumented immigrants in the United States states, 1998—2005 (Order No. 3284405). Available from ProQuest Dissertations & Theses A&I. (304889443). Retrieved from https://search-proquest-com-s.web.bisu.edu.cn/docview/304889443?accountid=41515

76.

U.S. Department of Education, National Center for Educational Statistics. (2014). The condition of education 2014 (Status dropout rates, NCES 2014-083). Retrieved from https://nces.ed.gov/pubs2014/2014083.pdf

77.

U.S. Department of Education, National Center for Educational Statistics. (2015). The condition of education 2015 (Public school expenditures, NCES 2015-144). Retrieved from https://nces.ed.gov/pubs2015/2015144.pdf

78.

U.S. Department of Labor. (n.d.). Changes basic minimum wages non-farm employment under state law: Selected years 1968 to 2013. Retrieved from http://www.dol.gov/whd/state/stateMinWageHis.htm

79.

Vargas

E. D.

(2011). In-state tuition policies for undocumented youth. Harvard Journal of Hispanic Policy, 23, 43–58.

80.

Yoshikawa

(2010). Immigrants raising citizens: Undocumented parents and their young children. New York, NY: Russell Sage.

81.

Zavodny

(1999). Determinants of recent immigrants’ locational choices. International Migration Review, 33, 1014–1030.