Population Research and Policy Review

, Volume 33, Issue 1, pp 13–30

In-State College Tuition Policies for Undocumented Immigrants: Implications for High School Enrollment Among Non-citizen Mexican Youth

Authors

    • Labor and Population DivisionRAND Corporation
  • Trey Miller
    • Labor and Population DivisionRAND Corporation
Article

DOI: 10.1007/s11113-013-9307-4

Cite this article as:
Bozick, R. & Miller, T. Popul Res Policy Rev (2014) 33: 13. doi:10.1007/s11113-013-9307-4

Abstract

This paper examines the secondary effects of policies that extend or deny in-state tuition to children of undocumented immigrants. Drawing upon repeated cross-sections of 15–17-year-olds in the Current Population Survey across 1997–2010, we assess changes in high school enrollment rates among Mexican-born non-citizen youth—a proxy for the undocumented youth population. We find that Mexican-born non-citizen youth living in states that deny in-state tuition benefits to undocumented youth are 49 % less likely to be enrolled in school than their peers living in states with no explicit policy. Conversely, Mexican-born non-citizen youth living in states that grant in-state tuition benefits to undocumented youth are 65 % more likely to be enrolled in school than their peers living in states with no explicit policy. The enactment of these policies is unrelated to changes in school enrollment among naturalized citizens. Our findings lend support to the proposition that that the implementation of in-state tuition policies sends signals to immigrant youth about their future educational possibilities in the long-term, which in turn influences the extent to which they engage in school in the short-term.

Keywords

ImmigrantUndocumented immigrantSchool enrollmentState policies

For most youth in the United States, the transition to adulthood is launched via the completion of high school and subsequent enrollment in college (Aud et al. 2011). Driving this modal life course trajectory is an economy that increasingly requires a college degree for stable economic footing in adulthood, buttressed by a rapid expansion of postsecondary opportunities made available over the past few decades (Bozick 2008). Not all youth, however, are legally able to take advantage of these opportunities—namely those whose parents are undocumented immigrants. These youth navigate this transitional period in the life course within a policy context that is in flux, with no national policy unequivocally assuring their access to college and with states varying in their extension of in-state tuition based on parental citizenship.

Currently, 11 states offer in-state tuition to children whose parents are undocumented immigrants, and eight states deny in-state tuition to children whose parents are undocumented immigrants, herein referred to as “undocumented youth” or “undocumented students” for ease of expression.1 The implications of these policies are just beginning to receive attention by researchers. To date, there is some evidence that the enactment of policies extending in-state tuition is associated with an increase in college enrollment among foreign-born non-citizens (Kaushal 2008; Flores 2009) while the enactment of policies denying in-state tuition is associated with a decrease in college enrollment among foreign-born non-citizens (Bozick 2008). The present study builds on this nascent area of inquiry to explore whether higher education policies can also influence educational investments earlier in the life course—in particular, enrollment in high school.

In this study, we directly test whether accommodating policies affect enrollment decisions of undocumented high school students. Specifically, we use 14 years of national-level data to compare rates of school enrollment among Mexican-born non-citizen youth ages 15–17 before and after the enactment of state policies regarding the extension or denial of in-state tuition benefits to undocumented students. The results provide support to the hypothesis that higher education policies have “trickle down” effects that can alter schooling trajectories for a particularly vulnerable population of youth.

Background

At the core of the American dream is the idea that anyone, regardless of background, can achieve success through hard work. In the country’s infancy, this ethos helped spur massive waves of immigration, inviting entrants willing to sacrifice and persevere with a promise of opportunity. At the time, open immigration policies and public education were dually embraced, as they facilitated industrialization and urban development (National Research Council 1997). Presently, however, these elements—open immigration policies and public education—once celebrated as hallmarks of “American exceptionalism,” face serious social and political challenges. With large numbers of undocumented immigrants crossing permeable borders and with fears of internal job competition and terrorism, Americans are less supportive of illegal immigration than they have ever been in the past (Segovia and Defever 2010). As of January 2010, the Department of Homeland Security estimated that 1.2 million undocumented immigrants under the age of 18 were residing in the United States (Hoefer et al. 2010). Their educational and occupational futures remain uncertain, as public policies regarding their legal participation in key institutions of domestic life—including schools, the health care system, and the labor force—are not entirely clear.

With respect to schooling, a number of judicial and federal directives have set the context for the current scenario. In 1982, the Supreme Court case Plyer v. Doe, guaranteed all children in the United States, regardless of legal status, a free public education from kindergarten through the 12th grade. The provisions of this court decision, however, did not extend to postsecondary education, either for citizens or non-citizens. Uncertainty regarding postsecondary access for non-citizens was partially clarified in the Illegal Immigration Reform and Immigrant Responsibility Act of 1996 (IIRIRA), which prohibited states from granting in-state tuition to undocumented students unless any out-of-state legal citizen was extended the same benefit. Undocumented youth were still allowed to enroll in college, though typically classified as international students and charged up to seven times the amount of in-state tuition rates (Abrego and Gonzales 2010).

To provide conditional access to postsecondary education to undocumented youth, bills such as the Development, Relief, and Education for Alien Minors Act (or “DREAM Act”), which specify criteria through which undocumented youth can obtain permanent resident status and thus, legally work or enroll in college, have been proposed in Congress with varying stipulations since 2001. To date, though, the DREAM Act has failed to make it through both Congress and the Senate (Russell 2011). More recently, in June 2012, President Obama announced that the Department of Homeland Security would temporarily halt the deportment of undocumented youth on the condition, among others, that they are currently enrolled in school or received a high school diploma/GED certificate. By doing so, Obama explicitly tied the fate of these youth to their participation in the school system.

While at the federal level movement on the passage of the DREAM Act and related policy discussions have ebbed and flowed, a number of states have been proactive and explicit in their support of higher education to undocumented youth. In 2001, Texas became the first state to pass legislation granting undocumented students eligibility for in-state tuition, with California following suit in 2002. The seminal policy decisions of these two states were non-trivial, as approximately 40 % of all undocumented immigrants reside within their borders (Hoefer et al. 2010). After Texas and California, 11 additional states enacted legislation to extend in-state tuition benefits to undocumented youth: Connecticut, Illinois, Kansas, Nebraska, New Mexico, New York, Oklahoma, Rhode Island, Utah, Washington, and Wisconsin. In our study, we refer to these as states with accommodating policies. Table 1 lists these states, along with the month and year their policies became effective.
Table 1

State tuition policies for undocumented immigrants 1997–2012

State

Legislation

Date enacted

Policy allows in-state tuition: “Accommodating”

Policy denies in-state tuition: “Restrictive”

Alabamaa, b

HB56

September 2011

 

X

Arizona

Proposition 300

December 2006

 

X

California

AB540

January 2002

X

 

Colorado

HB 1023

August 2006

 

X

Connecticuta

HB 6390

July 2011

X

 

Georgia

SB 492

July 2008

 

X

Illinois

HB60

May 2003

X

 

Indianaa

HB 1402

July 2011

 

X

Kansas

HB 2145

July 2004

X

 

Nebraska

LB 239

July 2006

X

 

New Mexico

SB 582

April 2005

X

 

New York

SB 7784

August 2003

X

 

Oklahoma

SB 596

August 2003

X

 

Oklahoma

HB 1804

November 2007

 

X

Rhode Islanda, c

September 2012

X

 

South Carolinab

HB 4400

June 2008

 

X

Texas

HB 1403

June 2001

X

 

Utah

HB144

July 2002

X

 

Washington

HB 1079

July 2003

X

 

Wisconsin

A75

June 2009

X

 

Wisconsina

A75 revoke

July 2011

 

X

aPolicy enacted alter October 2010 and not included in the present analysis

bPolicies in Alabama and South Carolina bar undocumented immigrants from college enrollment altogether

cNot a state bill. Passed by Rhode Island’s Board of Regents

Not all states, however, have welcomed undocumented students into their system of higher education: Alabama, Arizona, Colorado, Georgia, Indiana, and South Carolina each passed legislation that restrict in-state tuition benefits to undocumented students, also shown in Table 1. In our study, we refer to these as states with restrictive policies. Changes in elected officials and in the social climate have led two states to change course from accommodating to restrictive: Oklahoma’s 2003 senate bill permitting both in-state tuition and financial aid to undocumented youth was overturned in 2007, and Wisconsin’s 2009 assembly bill permitting in-state tuition to undocumented youth was revoked in 2011. Both of these states now restrict tuition benefits to undocumented youth. The remaining states not listed in Table 1 do not have an explicit policy on the issue.

Studying the effects of these policies at the national level is difficult, as government-sponsored national surveys are not permitted to directly ask respondents their naturalization status for legal and ethical reasons, and studies that track the schooling careers of immigrant youth (such as the Children of Immigrants Longitudinal Study and the Longitudinal Immigrant Student Adaptation Study) lack state and time variability required to examine policies that are enacted in a staggered way across the country.2 These challenges, notwithstanding, a handful of studies have made use of nativity and citizenship status collected in the U.S. Census Bureau’s Current Population Survey (Bozick et al. 2013; Flores 2009; Kaushal 2008) and American Community Survey (Chin and Juhn 2011) to compare rates of college enrollment before and after the enactment of in-state tuition policies among respondents who self-identify as Mexican/Hispanic non-citizens. The rationale being that Mexicans comprise the lion’s share of the undocumented population, and so honing in exclusively on Mexicans/Hispanic immigrants who lack citizenship should yield the most valid estimates of the policy’s true effect. The findings are mixed, with two of the studies showing that the extension of in-state tuition rates to undocumented youth was associated with higher rates of college enrollment among those who identify as foreign-born non-citizens (Flores 2009; Kaushal 2008) and two of the studies yielding null findings ( Bozick et al. 2013; Chin and Juhn 2011). More recent work by Bozick et al. (2013) finds a decline in college enrollment among Mexican-born non-citizen youth in states that enact restrictive policies. Taken as a whole, these studies highlight the role that state policies might potentially play in providing opportunities for human capital development for immigrant youth, as well as potential methodological approaches that can be used when studying the undocumented population using national-level data sources.

Motivation for the Present Study

Our study builds on this body of research by examining whether the enactment of in-state tuition policies, which are aimed explicitly at higher education, also influence pre-college preparation among immigrant youth. Because past studies focus on the years immediately after the policies were enacted, they are unable to capture the effect of accommodating tuition policies on students who would have enrolled in college if they had sufficient time to alter their college preparation behavior during high school. By explicitly examining the short-term effect of these policies on pre-college preparation, we are able to assess whether accommodating tuition policies are likely to have larger effects on college-going rates than those reported in Kaushal (2008) and Flores (2009) in the long run.

More broadly, our inquiry is part of a nascent body of research that finds that the expansion of postsecondary opportunities can encourage academic investments among students while they are still in high school. Youth are aware of opportunities in their broader communities, and adjust their expectations and their behaviors accordingly to align with such opportunities (Morgan 1998). For example, Henry and Rubenstein (2002) found that Georgia’s helping outstanding pupils educationally (HOPE) scholarship program, which provides full tuition to Georgia’s public colleges for state residents with at least a 3.0 GPA, led to increased grades and SAT scores among minority youth in the state. Similarly, Domina (2007) showed that following the introduction of scholarship programs in Texas that provide tuition subsidies to economically disadvantaged youth there was a decrease in truancy and an increase in advanced math course taking at high schools whose students qualified for the award. More recently, Bartik and Lachowska (2012) found an increase in credit accrual and a decrease in suspensions among high school students following the implementation of a college scholarship program in Kalamazaoo, Michigan. Taken together, these studies demonstrate that programs that broaden postsecondary opportunities to a broader swath of the student population can have positive effects that “trickle down” to the high school arena.

Though not focused directly on higher education, there is some research showing that immigrant youth’s schooling experiences are shaped by state policies toward immigrant incorporation. Filindra et al. (2011) find that immigrant youth attending high schools in states that extend generous welfare benefits (TANF, Medicaid, Supplemental Security Income, food stamps, etc.) to immigrants and refugees were more likely to graduate from high school than their peers living in states with less generous benefits. In interpreting this finding, the authors speculated that immigrants trust American institutions more when they are made to feel welcome, and hence their children develop positive attitudes toward their new schools. The logic here accords with the aforementioned studies on policies that support postsecondary enrollment: states that provide opportunities for upward mobility may encourage immigrant youth to invest more in their education.

The present study builds on these findings by assessing whether the extension of in-state tuition benefits to undocumented youth influences high school enrollment among Mexican immigrants living in the United States. To do so, we emulate the approach taken by Kaushal (2008), Flores (2009), and Bozick et al. (2013), and use foreign-born non-citizen youth in the Current Population Survey as a proxy sample for the undocumented youth population. Without access to in-state benefits, undocumented youth are forced to pay out-of-state tuition rates—a substantial barrier for youth from Mexican immigrant families, many of whom have limited economic resources (Abrego and Gonzales 2010). When accommodating states shown in Table 1 initially adopted these policies, they were in effect providing a new form of financial support that was previously unavailable to undocumented immigrant youth, thus signaling new opportunities to them about their futures. Therefore, in accord with the aforementioned research, we anticipate that rates of high school enrollment will increase (decrease) among Mexican-born non-citizen youth following the passage of accommodating (restrictive) in-state tuition policies. Without tuition benefits, the motivation for undocumented youth to remain in school might be diminished. Instead, it may make more sense for them to leave school to gain traction in the labor force as a means to lay the foundation for their occupational futures. In examining these possible relationships, our study provides a contribution to the burgeoning literature on how a uniquely disadvantaged segment of the youth population responds to changes in policies that expand or contract their opportunities after high school.

Though our study cannot disentangle the complex trade-offs under-girding migration and school enrollment decisions, our focus on non-citizen Mexican youth complements recent research that highlights how both sets of decisions may jointly reflect cultural orientations toward work among recent Mexican migrants, many of whom are undocumented (Kandel and Massey 2002). For example, Oropesa and Landale (2009) find that youth who recently migrated from Mexico are unlikely to enroll in school as they left their native communities primarily to pursue labor market opportunities in the United States. Those youth who do enroll in school are more likely to drop out the longer they remain in the United States (Oropesa and Landale 2009), with many leaving school to enter the labor market (Bradley and Renzulli 2011; Olatunji 2005). Bachmeier and Bean (2011) contend that recent Mexican immigrant youth face a number of obstacles to success as they acculturate to their new communities, and so they invest heavily in either school or work (but not both) as a resilience mechanism to develop human capital valued in the United States. This pattern of school or work specialization is most pronounced for recent migrants. Taken together, these findings suggest that policies that deny in-state tuition might impel children of recent migrants from Mexico to leave school and establish themselves in the labor force should they perceive structural barriers to their education (Bachmeier and Bean 2011).

Data

To address whether changes in state tuition policies influence high school enrollment rates of undocumented youth, we analyze the Current Population Survey (CPS), a nationally representative monthly sample survey of approximately 60,000 households in 792 sample areas across the country that provides a wide range of information on population characteristics and the state of the labor force. The U.S. Census Bureau, using a scientifically selected national sample of residential housing units with coverage in all 50 states and the District of Columbia, conducts the survey for the U.S. Bureau of Labor Statistics. For our analysis we use a pooled sample of 64,578 youth between the ages of 15 and 17 who were included in the CPS October household interview from 1997 to 2010. We use the October interview because it is the only month in which the CPS includes questions that determine the school enrollment of every person (over the age of three) living in the residential unit. We use 1997 as a starting point as it corresponds to the passage of IIRIRA and provides 4 years of “pre-policy” baseline data prior to 2001, the first year that a state (Texas) adopted an accommodating tuition policy.

Using information on their nativity and citizenship we classify sample members as either citizens (either foreign-born or native-born) or foreign-born non-citizens. Following the methodological approach of Kaushal (2008), Flores (2009), and Authors (2013), we use foreign-born non-citizens as a proxy for undocumented immigrants as this classification provides the most accurate delineation possible within the CPS data. Our primary analytical focus is on a subsample of 1,379 non-citizen youth born in Mexico, as these youth have the highest probability of being children of undocumented immigrants and tend to draw the most attention from policy makers in discussions regarding pathways to citizenship and/or deportation.3 Moreover, foreign-born non-citizens from countries other than Mexico are more likely to hold legal authorization to be in the United States and are thus less likely to be affected by the polices that we study. Including them in our proxy group for undocumented immigrants risks underestimating the effect of policies on school enrollment. For comparative purposes, however, we also examine the high school enrollment patterns of three groups of citizens: 3,311 Mexican youths, 5,682 non-Mexican Hispanic youths, and 54,206 White youths (who comprise the majority population in the United States during our period of study).

The dependent variable in our analysis is school enrollment. For each member of our analytic sample we determine whether or not they were enrolled in school at the time of the October interview. This binary outcome is coded “1” if enrolled and “0” if not enrolled. Across the 14 years of our study, rates of school enrollment were 85.4 % for Mexican-born non-citizens, 95.6 % for Mexican youth, 95.2 % for non-Mexican Hispanic youth, and 97.2 % of White youth. Note that rates of school enrollment in the CPS are relatively high. This is in part by design, as October represents the month when enrollment in school is at its highest, before most incidences of within academic calendar year dropping out occur. Additionally, the CPS only includes the non-institutionalized population. Vulnerable populations that traditionally have low rates of school enrollment—such as youth in hospitals or mental health facilities, youth in jails or detention centers, and youth living on the street or in homeless shelters—are excluded from the CPS sample. This design limitation of the CPS in fact sharpens our estimation of the relationship between policy changes and school enrollment, as our focus is on voluntary withdrawal from or persistence in school.

In our multivariate analyses, we include a set of control variables that are strongly associated with educational outcomes: sex, age, family structure, annual family income, time of migration, household language, metropolitan area residence, state compulsory school attendance laws, and the local unemployment rate. As a population-based study primarily intended to measure labor force participation, the CPS is limited in the number of variables available to control for ethnic and/or migration status differences in schooling experiences. The control variables used in our analysis capture an array of factors that are known to influence the educational attainment of immigrants (Portes and Rumbaut 2001), and are commensurate with other studies that use Census data to construct multivariate models of school enrollment of immigrant youth (Hirschman 2001; Oropesa and Landale 2009). The distribution for each of the control variables by our key citizenship groups is shown in Table 2.
Table 2

Distribution of control variables by citizenship and ethnic origin

 

Mexican-born non-citizens

US citizens

Mexican

Hispanic, non-Mexican

White

Sex

 Female

48.7 %

49.8 %

49.3 %

48.8 %

 Male

51.3 %

50.2 %

50.7 %

51.2 %

Age

 15

30.2 %

36.4 %

35.0 %

34.1 %

 16

34.2 %

34.0 %

34.1 %

33.9 %

 17

35.6 %

29.6 %

30.9 %

32.0 %

Living with two biological parents

 Yes

68.5 %

68.7 %

56.3 %

74.7 %

 No

31.5 %

31.3 %

43.7 %

25.3 %

Family income (ln)

9.94

10.22

10.38

10.89

Immigrated to US in the past 5 years

 Yes

30.3 %

0.9 %

1.0 %

0.1 %

 No

69.7 %

99.1 %

99.0 %

99.9 %

Primary household language

 Spanish

43.4 %

17.2 %

5.9 %

0.3 %

 Non-Spanish

56.6 %

82.8 %

94.1 %

99.7 %

Metropolitan area residence

 Yes

88.6 %

88.5 %

86.9 %

71.0 %

 No

11.4 %

11.5 %

13.1 %

29.0 %

Age of compulsory school attendance

 16

29.6 %

20.3 %

42.9 %

54.8 %

 17

24.1 %

24.8 %

22.2 %

16.8 %

 18

46.3 %

54.9 %

34.9 %

28.4 %

Local unemployment rate

6.2 %

6.7 %

6.0 %

5.2 %

N

1,379

3,311

5,682

54,206

Sex is a binary variable coded “1” if the student is male and “0” if the student is female. Age is represented by two binary variables, one coded “1” if the sample member is 17 years old and the other coded “1” if the sample member is 16 years old. Sample members who are 15 years old serve as the reference category. Family structure is binary variable coded “1” if the sample member lives with their two biological parents and “0” otherwise. Annual family income is a continuous measure created from a categorical measure available (e.g., less than $5,000; $5,000 to $9,999; $10,000 to $14,999; etc.). We assigned the midpoint for each category and used the Pareto distribution to assign the value to the top open-ended category ($200,000 and over). We then transformed this continuous measure to its natural logarithm. Time of migration is a binary variable coded “1” if the sample member migrated to the United States within 5 years of the October interview, and “0” if the sample member migrated to the United States more than 5 years prior to the October interview or if they were born in the United States. Household language is measured by a binary variable coded “1” if Spanish was the primary language spoken in the sample member’s household, else “0.” Metropolitan area residence is a binary variable coded “1” if the sample member lived in metropolitan statistical area (MSA) as defined by the US Census Bureau.

Lastly, we control for two contextual variables that may be particularly important in signaling sanctions and opportunities to youth when making decisions regarding staying in school: state compulsory school attendance laws and the local unemployment rate. State compulsory attendance laws are taken from the National Center for Education Statistics and are represented by two time-varying binary variables, one coded “1” if the sample member lives in a state that requires attendance until age 18 and the other coded “1” if the sample member lives in a state that requires attendance until age 17. Youth living in states that require attendance until age 16 serve as the reference category. The local unemployment rate is a continuous time-varying measure provided by the Bureau of Labor Statistics that indicates the state, county, or MSA level unemployment rate depending on the most proximal level of aggregation available in the public-use CPS files.4

Analytic Strategy

We assess the effects of state tuition policies by estimating the following model:
$$ Y_{ist} = \gamma_{1} ACCOMODATE_{st} + \gamma_{2} RESTRICT_{st} + \beta X_{i} + \delta_{st} + \tau_{st} + \upsilon_{s} + \upsilon_{t} $$

In the model, Y is the predicted probability that youth i, in state s, and year t (1997–2010) is enrolled in school. For ease of interpretation we present the model as linear, but we estimate it using a standard logit model. Specifically, using a maximum likelihood logit link function, we estimate Y as a function of two mutually exclusive policy status variables ACCOMMODATE and RESTRICT that correspond to the state policies listed in Table 1. The reference category in our model consists of sample members living in states in years in which there was no explicit tuition policy in place. Our two parameters of interest are γ1 and γ2: γ1 is the parameter estimate corresponding with ACCOMODATE, which is a binary variable coded “1” in the years where the state had a policy granting in-state tuition to undocumented immigrants (else “0”), and γ2 is the parameter estimate corresponding with RESTRICT, which is a binary variable coded “1” in the years where the state had a policy restricting undocumented immigrants from receiving in-state tuition (else “0”). Also included in the model is X, a vector of control variables (sex, age, family structure, annual family income, time of migration, household language, and metropolitan area residence) measured at the individual level i; δ is the age of compulsory school attendance that varies by state s and year t; τ is the rate of unemployment that varies by state s (or county or MSA) and year t; υs are state fixed-effects; and υt are year fixed-effects. In the model estimation, the standard errors are adjusted for clustering at the state level.

Crucial to our identification strategy is the inclusion of state fixed-effects (υs). It is likely that states that adopt accommodating or restrictive policies differ systematically from other states in ways that may influence school enrollment rates of the foreign-born. For example, states with more accommodating tuition policies may be more accepting of undocumented immigrants in general, which may in turn lead to more welcoming and culturally sensitive schools, and hence, high rates of enrollment among the undocumented—even in the absence of an accommodating policy. On the other hand, states may adopt accommodating policies as a means to strengthen their labor force because they have large numbers of undocumented immigrants with low levels of educational attainment who are likely to stay in the state for the foreseeable future. The reasons for adopting accommodating or stringent restrictive may differ significantly across states in ways that are related to a state’s baseline school enrollment rate. By including state fixed-effects, υs, we are able to remove the potentially confounding influence of all state-level factors that remain constant over time.

If the enactment of policies granting or restricting tuition to undocumented immigrants have a causal effect on the school enrollment of undocumented students of high school age, then we expect γ1 and γ2 to yield significant coefficients when the model is estimated on our sample of Mexican-born non-citizen youth. As citizens are guaranteed in-state tuition, their rates of enrollment should be unaffected by policy changes that have no bearing on their educational opportunities. Therefore, we will re-estimate the model for our sample of three citizen groups (Mexican youth, non-Mexican Hispanic youth, and White youth) as a falsification test on the viability of our findings. Should we find γ1 and γ2 to be significant in our models based on the three citizen samples, this would suggest that there were time-varying unobserved factors correlated with the enactment of in-state tuition policies that affected school enrollment of all youth, regardless of whether they were undocumented immigrants. On the other hand, should we find γ1 and γ2 to be non-significant in our models based on the three citizen samples (which in theory should contain zero undocumented immigrants) and significant in our sample of Mexican-born non-citizen youth (which should be comprised largely of undocumented immigrants), we will have stronger evidence that the policy is directly affecting school enrollment among undocumented immigrants.5

Findings

To provide context to our multivariate model, we first examine unadjusted rates of school enrollment for our sample of Mexican-born non-citizens contingent on the policy status of their state: restrictive policy, no policy, or an accommodating policy. We pooled across all 14 years of data and calculated the proportion enrolled in school. These estimates are shown in Fig. 1. In accord with our theoretical predictions, Mexican-born non-citizen youth are most likely to enroll in school when living in states that extend tuition benefits to undocumented students and are least likely to enroll in school when living in states that deny tuition benefits to undocumented students. The difference in rates of enrollment between restrictive and accommodating states is substantial: 78.6 % of Mexican-born non-citizen youth living in restrictive states are enrolled in school compared with 90.9 % of Mexican-born non-citizen youth living in accommodating states. This pattern does not hold among Mexican-born citizens, and though somewhat evident for non-Mexican Hispanic and White youth, it is far less pronounced. These descriptive comparisons provide tentative support for the hypothesis that policies that provide access to higher education can support a commitment to schooling at younger ages and conversely, policies that restrict access to higher education can attenuate a commitment to schooling at younger ages.
https://static-content.springer.com/image/art%3A10.1007%2Fs11113-013-9307-4/MediaObjects/11113_2013_9307_Fig1_HTML.gif
Fig. 1

Proportion enrolled in school by in-state tuition policy status

To provide a more rigorous test of the hypothesis, we estimated the logit model described earlier separately for each of our four samples. We converted the log odds coefficients into odds ratios for ease of interpretation, which we show in Table 3. Odds ratios greater than 1.00 indicate a higher odds of school enrollment relative to the reference category and odds ratios less than 1.00 indicate a lower odds of school enrollment relative to the reference category. We suppress the estimates for the state and year fixed-effects for clarity of presentation as they are used simply as controls and are not of substantive interest.
Table 3

Odds ratios from logit regression models predicting school enrollment among 15–17-year-olds

 

Mexican-born non-citizens

US citizens

Mexican

Hispanic, non-Mexican

White

Sex

 Female (reference)

1.00

1.00

1.00

1.00

 Male

0.79

0.76

1.18

0.82**

Age

 15 (reference)

1.00

1.00

1.00

1.00

 16

0.47*

0.33**

0.50**

0.44**

 17

0.24**

0.15**

0.24**

0.25**

Living with two biological parents

 Yes

2.02**

1.30

1.23

1.18

 No (reference)

1.00

1.00

1.00

1.00

Family income (ln)

1.11

1.34**

1.69**

2.12**

Immigrated to US in the past 5 years

 Yes

0.26**

2.13

2.23

1.55

 No (reference)

1.00

1.00

1.00

1.00

Primary household language

 Spanish

0.91

0.73

0.81

1.00

 Non-Spanish (reference)

1.00

1.00

1.00

1.00

Metropolitan area residence

 Yes

0.98

0.61

0.74

0.94

 No (reference)

1.00

1.00

1.00

1.00

Age of compulsory school attendance

 16 (reference)

1.00

1.00

1.00

1.00

 17

1.06

2.62

1.91**

0.63

 18

1.42

2.78*

1.23

0.94

 Local unemployment rate

1.09

1.25**

1.10

1.02

In-state tuition

 Restrictive policy

0.51**

2.22

0.76

0.93

 No policy (reference)

1.00

1.00

1.00

1.00

 Accommodating policy

1.65*

0.95

0.83

0.98

N

1,379

3,311

5,682

54,206

All models control for state and year fixed-effects

* p < 0.05; ** p < 0.01

Though we are primarily interested in the parameters that correspond with our state policy variables, there are two findings among the control variables of note. First, the only set of parameters that yield significant point estimates across all four samples are those for age: 16- and 17-year-olds are significantly less likely to be in school than their 15-year-old counterparts, regardless of citizenship or ethnic origin. This reflects the increasing number of youth who drop out of school as they enter the later years of high school, particularly as they reach and surpass state mandatory age requirements for school enrollment. Second, among Mexican-born non-citizens, the timing of immigration is strongly related with school enrollment: those who immigrated to the US within the past 5 years are 74 % less likely to be enrolled in school than their peers who immigrated to the US more than 5 years ago. This accords with past research which finds that the longer that immigrant youth from Mexico live in the United States, the more likely they are to enroll in school (Hirschman 2001).

Turning to our key parameters of interest, we find that both γ1ACCOMMODATE and γ2RESTRICT yield significant coefficients among our sample of Mexican-born non-citizen youth. Specifically, Mexican-born non-citizen youth living in states that deny undocumented youth access to in-state tuition benefits are 49 % less likely to be enrolled in school than their peers living in states with no explicit policy. Conversely, Mexican-born non-citizen youth living in states that grant in-state tuition benefits to undocumented youth are 65 % more likely to be enrolled in school than their peers living in states with no explicit policy.6 These patterns accord with those detected in Fig. 1 and support the contention that immigrant youths’ investments in schooling are contingent upon their access to higher education.

As a falsification test, we examined whether these relationships were also present among the other three citizen groups. Across these three samples, we found that the enactment of restrictive or accommodating policies was unrelated to school enrollment—suggesting that the estimated effects of tuition policies among Mexican-born non-citizen youth were not due to unobserved time-varying factors correlated with state-level legislation.7 We cannot with certainty rule out the threat of spurious findings due to unobserved factors. However, that we detect significant effects among our key policy variables net of individual, state, and year characteristics, and that we do not detect similar findings among non-targeted groups lends support to the hypothesis that policies expanding or restricting opportunities for higher education influence academic engagement at earlier stages of schooling.

Discussion

Immigrant youth in the United States occupy a precarious position within the school system. They are ensured primary and secondary education irrespective of their legal status, but their access to college is contingent on both their legal status and their state of residence. Currently, 11 states offer in-state tuition to undocumented youth (e.g., accommodating policies) and eight states deny in-state tuition to undocumented youth (e.g., restrictive policies). The consequences of these policies are not benign, as the acquisition of a postsecondary degree is critical in helping to lay the foundation for later occupational and economic attainment. Because schools are intended to be the central institution through which immigrant youth acculturate to their host country and acquire critical forms of human capital needed to succeed in the economy, blocked opportunities may preclude solid footing on the “ladder of social mobility” that helped support generations of immigrants before them. To inform ongoing policy discussions regarding the educational fate of youth whose parents illegally immigrated to the United States, our study helps to address the question: do state policies regarding postsecondary access influence patterns of human capital acquisition among undocumented immigrants?

In examining changes in state policies from 1997 to 2010, our study provides evidence that Mexican-born non-citizen youth between the ages of 15 and 17—a group most likely to be comprised of undocumented immigrants—had the highest odds of school enrollment when living in states with accommodating tuition policies. Earlier research shows that college enrollment rates of foreign-born non-citizen youth increase once states extend tuition benefits (Flores 2009; Kaushal 2008). Our analysis finds that these policies may also influence school enrollment prior to the college years. Specifically, we found that Mexican-born non-citizen youth living in states that grant in-state tuition benefits are 65 % more likely to be enrolled in school than their peers living in states with no explicit policy. Like recent research which shows a decline in college enrollment for undocumented youth in states that enact restrictive policies (Bozick et al. 2013), we find that these policies may also limit human capital acquisition at earlier stages of schooling: Mexican-born non-citizen youth living in states that restrict their access to in-state tuition benefits are 49 % less likely to be enrolled in school than their peers living in states with no explicit policy.

More broadly, our findings accord with research on other college access programs targeted to disadvantaged groups which by and large find that when opportunities for postsecondary education are expanded, high school-aged youth will invest more in their schooling (Bartik and Lachowska 2012; Domina 2007; Henry and Rubenstein 2002). These studies, along with ours, suggest that the implementation of accommodating policies sends signals to youth about their future educational possibilities in the long-term, which in turn influences the extent to which they engage in school in the short-term. However, as a population-based study we do not have information on youth’s perceptions of opportunities, their educational expectations, or their knowledge of different college subsidy policies and programs, and as such we can only provide indirect evidence about these underlying processes that drive enrollment decisions. More research is needed on how these policies are communicated to immigrants, who often are distrustful of government and/or lack full proficiency in English; how these policies complement or diverge with immigrant families’ expectations about work and school; and whether these investments in schooling are sustained in a fluid and potentially hostile policy environment. One promising avenue would be to explore how undocumented immigrants from other countries and racial/ethnic groups respond to these and similar policies. The convergence or divergence of such experiences with those of Mexican immigrants could illustrate the magnitude of policy effects apart from the conditions of reception and modes of incorporation that vary widely across immigrant groups.

Though the findings from our multivariate models are relatively robust, we caution readers of two methodological limitations when drawing conclusions. First, as an observational study, youth and their families in the CPS are not randomly assigned to legal status and/or states of residence. To attenuate the threat of spurious relationships, we employed a pre-/post-time-series design that controls for sociodemographic characteristics, compulsory school attendance laws, local unemployment, year fixed-effects, and state fixed-effects. However, there could still be unobserved factors simultaneously inducing changes in state policies and school enrollment and as such, our analysis provides suggestive, and not causal, evidence about the influence of in-state tuition policies. Second, given the challenges associated with identifying and studying undocumented immigrants in large-scale national surveys, we are only able to use a proxy category for our key target group. Not all Mexican-born non-citizens are undocumented immigrants, and as a consequence our findings only provide an approximation of the true relationship between in-state tuition policies and the school enrollment rates of undocumented youth. That said, we remind readers that there is no nationally representative individual-level data set that identifies undocumented respondents with certainty. Analyses of existing data sources such as the CPS are necessary to begin an assessment of these policies, as without them there is little evidence to bear on their efficacy.

In closing, as the country continues to grapple with the most effective means for dealing with the large flux of new entrants, immigrant youth—including those who were brought to the United States illegally by their parents—will continue to progress through a schooling system that sends them mixed signals regarding their long-term educational opportunities and consequently, their short-term need to invest in their studies. The findings from the present study indicate that immigrant youth are reactive to their opportunity structures, and that any attempt to expand or to restrict those opportunity structures should consider the secondary effects which could potentially attenuate if not halt the generational procession of upward mobility for immigrants. If schools remain the primary institution through which immigrant youth acculturate to the host society and acquire relevant forms of human capital, then with stark differences in state policies toward higher education access we may anticipate differential rates of upward mobility that vary geographically.

Footnotes
1

We consider undocumented youth as those who were born outside the United States to parents who are not authorized to reside or work in the United States. These youth were brought to the United States by their parents before they turned age 18. This does not include those whose parents are naturalized citizens, hold green cards, or hold any other form of legal authorization.

 
2

While our analysis and our review of past research is focused on the implementation of these policies at the national level, it is worth nothing that a handful of studies focusing only on Texas found that the extension of in-state tuition in 2001 was associated with increases in college enrollment within the state (Flores 2010; Dickson and Pender 2013).

 
3

As of 2010, immigrants from Mexico comprised 62 % of the undocumented population in the United States (Hoefer et al. 2010).

 
4

In our analytic sample, 37.4 % of sample members had unemployment rates measured at the state level, 7.5 % of sample members had unemployment rates measured at the county level, and 55.1 % of sample members had unemployment rates measured at the MSA level. We included a control variable in all models (parameter estimates not shown) indicating the geographic level used to measure unemployment.

 
5

It is also possible, albeit unlikely, that students other than undocumented immigrants would experience a “crowding out” effect from accommodating tuition policies. If those students anticipate increased competition from undocumented immigrants for college admission as a result of the policy, they may be less likely to enroll in college. Under this interpretation, a zero coefficient for γ1 and γ2 is indicative of a lack of evidence for such a “crowding out” hypothesis.

 
6

In supplementary analyses not shown, we replaced the binary policy indicators with continuous measures that indexed the number of months since each policy was enacted. These estimates mirrored those obtained using the binary indicators: each additional month since the enactment of a restrictive policy was associated with a reduction in the odds of school enrollment among Mexican-born non-citizen youth and each additional month since the enactment of an accommodating policy was associated with an increase in the odds of school enrollment among Mexican-born non-citizen youth. These findings are available upon request from the authors.

 
7

In analyses not shown, we also estimated the model for three other groups of non-native-born, non-citizens: Asians, non-Mexican Hispanics, and Black. Across these three groups, the enactment of restrictive or accommodating policies was unrelated with school enrollment. As these samples are less likely to contain to undocumented immigrants (and instead include legally authorized youth, exchange students, etc.), it is not clear whether there is in fact a null effect or if the samples are “underpowered” with undocumented youth to detect an effect. Additionally, there may be specific racial-dynamics at play in how these policies are communicated to and enforced among those perceived to be undocumented, the majority of whom are Mexican. Population-level analyses using data like the CPS are not equipped to detect such dynamics.

 

Acknowledgments

This research was supported by a generous grant to the authors from the Spencer Foundation (#201200043). All analyses and interpretations are the authors alone.

Copyright information

© Springer Science+Business Media Dordrecht 2013