Darolia & Potochnick / Educational Implications of IRT Policies 507

The Review of Higher Education

Summer 2015, Volume 38, No. 4, pp. 507–535
Copyright © 2015 Association for the Study of Higher Education
All Rights Reserved (ISSN 0162–5748)

Educational “When,” “Where,”

and “How” Implications of In-
State Resident Tuition Policies
for Latino Undocumented
Immigrants
Rajeev Darolia and Stephanie Potochnick
Difficult economic conditions faced by U.S. states and their residents have
drawn focus to the targeting of public expenditures and the efficient use of
funds to encourage economic growth. Concurrently, states are evaluating how
to best develop the productive resources in their communities by adopting
immigration and education related policies. A policy area that resides at the
nexus between these policy areas includes decisions about whether to use
state funds to encourage college matriculation and degree attainment among
undocumented immigrants. Between 2001 and 2013, 17 states adopted an
in-state resident tuition (IRT) policy that allows undocumented immigrants
to pay the relatively low tuition and fees available to citizen and legal per-
manent resident (LPR) state residents at public colleges.

Rajeev Darolia is an Assistant Professor in the Truman School of Public Affairs and College
of Education at the University of Missouri. Stephanie Potochnick is an Assistant Professor in
the Truman School of Public Affairs and public health program at the University of Missouri.
Authors are listed in alphabetical order and senior authorship is shared. We thank numerous
seminar participants for useful feedback and Claire Donze for her research assistance on this
project. Address queries to Rajeev Darolia, Truman School of Public Affairs, University of Mis-
souri-Columbia, 234 Middlebush Hall, Columbia, MO 65211–6100. daroliar@missouri.edu.
508 The Review of Higher Education Summer 2015

Research on IRT policies to date has mostly focused on the decision of

whether to enroll in college, and to a lesser extent, college completion. While
not uniform, most research indicates that by making higher education more
affordable, IRT policies positively affect college enrollment and associate
degree completion among Mexican and Latino foreign-born non-citizens
(FBNCs)—the strongest proxies for undocumented status (Conger, 2014;
Flores, 2010a; 2010b; Kaushal, 2008).1 Other research finds that IRT policies
also motivate high school youth to graduate with the hope to enroll in college
in the future (Bozick & Miller, 2014; Potochnick, 2014).
Financial and legal constraints associated with IRT policies, however, are
also likely to affect when and where students go to college, and how they pay.
While the size of the tuition discount associated with IRT policies is substan-
tial (average reduction for a 4-year institution in 2005 was $6,925; Kaushal,
2008), in-state tuition is still unaffordable for many undocumented young
adults, 40% of whom live below the federal poverty line (Gonzales, 2009).
Moreover, undocumented young adults are ineligible for federal financial aid
and state aid in most states. Even if undocumented young adults can afford
to attend college, upon graduation many of them find themselves in “legal
limbo,” unable to obtain work because of their immigration status (Suàrez-
Orozco, Yoshikawa, Teranishi, & Suàrez-Orozco, 2011). As a consequence,
undocumented immigrant youth may make educational investment deci-
sions that place high emphasis on limiting risk by focusing on short-term
costs and minimizing debt.
The decisions related to the type of educational investment undocumented
immigrant youth make, as well as how they finance their education, could
have a lasting impact on their economic future and the economic growth of
the state. Given that more than 11 million undocumented immigrants (30%
of the foreign-born population) live in the U.S. and more than 80,000 of them
reach college age each year (Passel & Cohn, 2008; Passel, 2003), understanding
the college behaviors of this population is important for policy development.
Encouraging postsecondary educational training is a key employment and
economic development policy tool for states. While individuals benefit from
the higher wages associated with some postsecondary training, states can
benefit from a range of social returns, including higher tax revenues, more
productive communities, and reductions in social service expenditures (Avery
& Turner, 2012; Moretti; 2004; Wolfe & Haveman, 2003). Research, however,
suggests that where students complete their post-secondary training (e.g.,
2-year vs. 4-year colleges) will affect the size of the educational returns (Kane

1
Chin and Juhn (2011) do not find strong evidence that IRT policies increase college
enrollment; they caution that their non-results may reflect small sample size challenges or
the shortness of the pre-post time period examined (2000–2005).
Darolia & Potochnick / Educational Implications of IRT Policies 509

& Rouse, 1995) and that the decisions students make during their educational
careers, including when to start, how to attend (e.g., full-time or part-time),
and how to finance college (e.g., via loans or work), can affect academic and
post-school outcomes (e.g., Light, 2001; O’Toole, Stratton, & Wetzel, 2003).
Using two nationally representative data sources, the Current Population
Survey (CPS) and the National Postsecondary Student Aid Study (NPSAS),
we are among the first to examine how IRT policies affect Latino FBNCs’
decisions on a range of intensive margin college decisions (i.e., when, where,
and how). We follow a research approach akin to a difference-in-differences-
in-differences method, in which we identify policy effects based on differences
in pre- and post-policy educational outcomes between Latino FBNCs cov-
ered and not covered by the policy, net of the educational trends of citizens,
and accounting for variation across states, over time, and within states over
time. For robustness, we compare Latino FBNCs to two citizen comparison
groups, Latino citizens and all citizens. We examine whether the adoption of
IRT policies affects when students enroll in college (e.g., how soon after high
school completion and age of enrollment) and where students enroll (e.g.,
2-year vs. 4-year and associate’s vs. bachelor’s degree). Further, we examine
how students finance their college education by assessing measures of student
borrowing, work behavior, and course-taking intensity.

States and IRT Policies

The debate over college access for undocumented immigrants began to
formalize in federal policy in 1996 with the Illegal Immigration Reform and
Immigrant Responsibility Act (IIRIRA), which prohibited states from provid-
ing in-state resident tuition benefits to undocumented immigrants unless
all U.S. citizens and nationals were eligible for the same benefits. Starting in
2001, however, 17 states have adopted (though two states later rescinded)
an IRT policy that lowers undocumented immigrant students’ costs associ-
ated with attending a public college2 and five states have reduced costs even
further by allowing undocumented immigrants access to state financial aid.
Moreover, the Board of Regents (or Governor) in three states has adopted
an internal policy similar to an IRT policy. In Table 1, we provide a list of
the 17 IRT policies.

2
Undocumented immigrants must meet certain residency requirements. While the crite-
ria varies for each state generally students must: 1) attend a school in the state for a certain
number of years; 2) graduate from high school in the state or receive a state issued GED; and
3) sign an affidavit stating that they have either applied to legalize their status or will do so
as soon as eligible (National Immigration Law Center [NILC], 2009).
 Table 1
510

Policy Provisions for States that Allow Undocumented Students to Gain Resident Tuition
Status as of 2013
State Date Date State Financial Legislation
Passed Enacted Aid for Undoc. Revoking Law: Residency Requirement: Years of High School in State
Date Effective Enacted

Texas 16-Jun-01 16-Jun-01 16-Jun-01 N/A Reside in-state with a parent 3-years prior to graduation and
graduate from a TX high school or GED program
California 12-Oct-01 1-Jan-02 1-Jan-13 N/A Attend a CA high school for 3 or more years prior to graduation
or GED
Utah 6-Mar-02 1-Jul-02 N/A N/A Attend a UT high school for 3 or more years prior to graduation
or GED
New Yorka 25-Jun-02 1-Aug-03 N/A N/A Two or more years at an approved NY high school, graduate from
NY HS or obtain a NY issued GED, and apply within 5 years
Washington 7-May-03 1-Jul-03 N/A N/A Complete a full senior year at a WA high school, live in WA at least
3 years immediately prior to diploma or equivalency
Oklahomab 12-May-03 12-May-03 12-May-03 1-Nov-07 Live in state with a parent or legal guardian for 2 years prior to
Revoked: Nov-07 graduation or GED
Illinois 18-May-03 20-May-03 1-Aug-11 N/A Attend IL high school for 3 years prior to graduation or GED and
reside with parent while attending IL high school
Kansas 20-May-04 1-Jul-04 N/A N/A Attend KS high school for 3 years prior to graduation or GED
New Mexico 5-Apr-05 5-Apr-05 5-Apr-05 N/A Attend NM high school for 1 year prior to graduation or GED
Nebraska 14-Apr-06 13-Jul-06 N/A N/A Reside in NB 3-years prior to graduation or GED and live with a
parent or guardian while attending high school
Wisconsin 26-Jun-09 29-Jun-09 N/A 26-Jun-11 Reside in WI 3 years prior to graduation or GED
Maryland 10-May-11 1-Jul-11 N/A N/A Attend MD high school for three years, prove parents filed taxes,
and for the first two years students can only attend
community colleges
Connecticut 1-Jun-11 1-Jul-11 N/A N/A Complete at least 4 years of high school level education in CT
The Review of Higher Education Summer 2015
Oregonc 2-Apr-13 1-Jul-13 N/A N/A Attend OR high school for three years, five years attendance in any
U.S.elementary or secondary school, and receive diploma in OR
within 3 years of enrolling in university
Coloradoc 29-Apr-13 29-Apr-13 N/A N/A Attend high school for at least 3 years prior to gradution or GED;
be admitted to a CO institution of higher learning within 12
months of graduating
Minnesotac 24-May-13 1-Jul-13 N/A N/A Attend MN high school for 3 years prior to graduation or GED
New Jerseyc 20-Dec-13 20-Dec-13 N/A N/A Attend NJ high school for 3 years prior to graduation or GED

Note. While Rhode Island, Hawaii, and Michigan have not adopted an IRT policy, the Board of Regents and Governors in these states have adopted a similar internal
policy. Sources: Flores (2007); NILC (2013); NCSL (2013)
a
Prior to NY’s policy, the State University of New York (SUNY) and the City University of New York (CUNY) provided in-state tuition to undocumented immigrants
except for during the spring of 2002
b
In 2007 OK passed another statute prohibiting undocumented immigrants from receiving in-state tuition but allowing the state’s Board of Regents (which wrote a
guideline memo in 2008) to award in-state tuition to undocumented students who attended an OK HS for at least two years. The legislation also made eligibility for
financial aid more restrictive.
c
Passed legislation outside of the sample period so treated as a non-policy state in analysis.
Darolia & Potochnick / Educational Implications of IRT Policies
511
512 The Review of Higher Education Summer 2015

The adoption of IRT policies has been highly contentious, with some states
eventually rescinding their policies and others facing legal challenges or im-
mediate counter-legislation. Aside from the political and legal debates, states
have incentives to encourage individuals to pursue postsecondary education
in-state. Research consistently finds positive and growing average private
returns to attending postsecondary education, including higher wages and
lower unemployment rates, even after taking into account growing college
costs (Avery & Turner, 2012). Importantly for policies that direct public
funding for higher education, college also produces average outcomes that
benefit society more broadly, with graduates associated with higher levels
of civic participation and charitable giving, less criminal activity, and more
productive communities among other social benefits (Moretti, 2004; Wolfe
& Haveman, 2003). Because public returns can exceed private returns, public
subsidies can ameliorate individual incentives to invest in one’s education at
a socially suboptimal level and therefore aid efficiency in the state economy
by increasing the supply of productive workers.
These rationales result in a variety of state supports for higher education.
While the federal government provides large financial aid programs such
as Pell Grants and subsidized student loans, states typically provide funds
that reduce public college tuition for residents of the state. IRT policies are
a form of state support that allows undocumented immigrants to pay the
lower in-state price. By reducing undocumented students’ tuition, these states
are agreeing to subsidize a portion of these students’ educational expenses
in an effort to encourage attendance at public colleges in the state. Extant
research indicates that lower tuition costs can lead to increased enrollment for
students generally (e.g., Dynarski, 2000; Heller, 1997) and that IRT policies
specifically can increase college attendance among undocumented students
(Flores, 2010a, 2010b; Kaushal, 2008).
For undocumented students, however, it is unlikely that postsecondary
attendance has the same public and private returns. Descriptive evidence
indicates that undocumented immigrants are often high-ability, but low-
income (Conger & Chellman, 2013), such that removing barriers to educa-
tion are likely to promote academic and vocational productivity. However,
since undocumented immigrants cannot legally work in the U.S. and face the
threat of deportation,3 expected private benefits that turn into public benefits,

3
While the 2012 Deferred Action for Child Hood Arrivals (DACA) executive order granted
undocumented immigrant youth a 2-year deportation waiver and work permit, it is unclear
whether these youth will be able to retain these work permits long-term or how employers
respond to these work permits. Moreover, the majority of IRT policies were adopted prior to
DACA, and thus, most undocumented immigrant youth in this study still faced deportation
and work limit challenges. While some undocumented immigrant youth are able to readjust
their status (Kaushal, 2008), most will remain unauthorized unless federal reforms are adopted.
Darolia & Potochnick / Educational Implications of IRT Policies 513

such as an augmented state tax base, might be limited by labor market bar-
riers. Moreover, in order to pay for their educational investments, students
may have to make sub-optimal finance and work decisions that hinder their
likelihood of realizing their investment and lower public and private returns.

Undocumented Immigrants and Educational Decisions

When making educational investment decisions, undocumented immi-
grants, like their authorized peers, are likely to weigh the present value of
expected benefits, such as higher expected wages and better labor market
outcomes, against the costs of such an investment, including direct tuition
costs and forgone earnings, as predicted by human capital theory (e.g.,
Becker, 1964). This comparison is far from straightforward for the average
student (who must be self-aware about their abilities and expected benefits
and costs), but is even more complex for undocumented students, who face
greater future uncertainty and potentially larger opportunity costs (e.g.,
choosing between paying for college and essential family needs; Greenman
& Hall, 2013).
Moreover, the unique social, community, economic, and policy contexts
facing undocumented immigrant youth shape the educational preferences
and informational knowledge that guide college choices. The literature on
college choice and immigrant assimilation emphasize the importance of
context in college decision making and demonstrate how policies of govern-
ments can alter the cost-benefit calculation by defining economic and social
opportunities (Perna, 2006; Portes & Rumbaut, 2006). These contextual
factors help us to understand the many potential routes to undocumented
immigrants’ college decisions, which may not appear to reflect a rational cost-
benefit decision since IRT policies do not reduce undocumented students’
uncertainty of the benefits (i.e., obtaining a college degree does not provide
permanent residency in the US or a legal right to work).
Uncertainty about future labor market prospects and residency in the U.S.
deters some undocumented immigrants from pursuing higher education
(Contreras, 2009). For others, though, a high level of resiliency, strong work
ethic, and optimism motivates them to pursue a college degree despite their
concerns about the future (Abrego, 2006, 2008; Contreras, 2009; Gonzales,
2009, 2012). An estimated 48% of the 65,000 undocumented students who
graduate from high school each year pursue some postsecondary education
(Passel, 2005), compared to 66% of high school graduates nationwide (U.S.
Bureau of Labor Statistics, 2013). The adoption of IRT policies has been
shown to push college enrollment levels among undocumented immigrants
even higher by up to four percentage points (Flores, 2010a, 2010b; Kaushal,
2008).
514 The Review of Higher Education Summer 2015

Qualitative evidence on IRT policies, however, suggests that undocu-

mented immigrants are making the decision to enroll in college with limited
information about college in general and the benefits for them in particular.
Because of their lower-economic status and fear of interacting with govern-
mental institutions, undocumented immigrant families often lack informa-
tion about the U.S. educational system and financial aid options (including
the availability of IRT policies in their state; Abrego 2006, 2008; Gonzales,
2008). As a consequence, undocumented Latino immigrant youth typically
have little institutional support and are forced to make investment decisions
(e.g., enrolling in 2-year colleges and attending part-time in order to work)
that have been shown to hinder college completion (Contreras, 2009; Perez &
Malagon, 2006; Perez & Rodriguez, 2011; Teranishi, Suarez-Orozco, & Suarez-
Orozco, 2011). For the few undocumented immigrant youth who graduate,
research suggest that they are surprised and frustrated that their legal status
forces them to return to the low-skilled jobs (e.g., cleaning offices, waiting
tables, and working in factories) they had hoped to escape (Gonzalez, 2009,
2012; Perez & Malagon, 2006).
Concurrent with the decision to enroll, students must make several
intensive margin decisions. One of the first is when to enroll. IRT policies
may shorten the amount of time that elapses between high school gradu-
ation and college enrollment for students who delay college entry in order
to save enough money to pay for college. Research suggests that delayed
entry may lower students’ future income according to a life-cycle model of
earnings, since entering school earlier can allow students to more quickly
accrue income gains associated with education (Ben-Porath, 1967). Delayed
entry has also been found to be associated with college dropout (Stratton,
O’Toole, & Wetzel, 2008), particularly among Latino students (Ganderton
& Santos, 1995). Research on Latino youth finds that they are more likely
to delay or pro-long their education (either by attending intermittently or
attending part-time) and as a consequence attend college into their mid to
late 20s—a time period when familial and work obligations can deter degree
completion (Fry, 2002; Ganderton & Santos, 1995; Hagy & Staniec, 2002)
A second decision students must make is where to attend. With regards
to 4-year vs. 2-year college choice, prior research on merit aid indicates that
lowering the cost of education can induce new students at the margin of
college entry to enroll in 2-year colleges and also push students from 2-year
colleges into 4-year colleges (Dynarski, 2004). For undocumented immi-
grants, there are reasons to believe that the price reduction associated with
IRT policies may have a larger effect on enrollment at 2-year colleges than
4-year colleges. Research indicates that the majority of Latino immigrants,
particularly Mexican immigrants, attend community colleges because they
cost less than 4-year colleges, offer more remedial coursework for English
Darolia & Potochnick / Educational Implications of IRT Policies 515

language learners, and are more accessible (geographically and education-

ally) and accommodating (particularly for working students; Fry, 2002,
Hagy & Staniec, 2002; Teranishi, Suarez-Orozco, & Suarez-Orozco, 2011).
These trends are supported based on research from Texas, which finds that
the state’s IRT policy has had the largest effect on enrollments at community
colleges (Texas Higher Education Coordinating Board, 2011) and less selec-
tive 4-year institutions (Dickson & Pender, 2013). While community colleges
can be a valuable resource for Latino immigrant youth, particularly those
seeking skills based education, the challenge is that many Latinos enroll in
2-year degrees with the intent to transfer but few end up doing so or even
completing their 2-year degree (Gonzales, 2008; Perez & Malagon, 2006;
Teranishi, Suarez-Orozco, Suarez-Orozco, 2011).
Lastly, students must determine how they will pay for their education.
These decisions have the potential to affect students’ patterns of enrollment,
as well as expected benefits associated with college. Because they have few
financial aid options and low familial resources, undocumented immigrants
may either use private loans or work to finance their education. If they work,
students may also have to adjust their enrollment intensity and only attend
part-time in order to balance work and school. Since IRT policies lower the
monetary investment associated with college enrollment, they may result in
some undocumented students having to borrow less or work fewer hours—
possibly enabling them to enroll full-time rather than part-time—in order
to pay for college costs. On the other hand, if the policies induce financially
constrained students to attend college, some of these students may not be
able to independently afford even the lower costs, and without access to
grants, these students may end up taking out burdensome student loan debt
or have to work while in school.
Research on Latinos and undocumented immigrants suggest that they
largely use work as a means to pay for their education and that the decision
to work can negatively affect their chances for success. Qualitative research
on undocumented immigrant college students find that many work burden-
some hours to meet their financial needs, which reduces their ability to attend
school full-time and to develop the supportive relationships that facilitate
college completion (Contreras, 2009; Gonzales, 2008; Perez & Malagon, 2006).
These trends are supported by quantitative assessments of undocumented
students in New York, which finds that the absence of IRT and financial aid
eligibility hinders students’ ability to enroll full-time (Conger, 2014; Conger,
& Chellman, 2013). Extant research on Latino college youth in general finds
that they struggle to balance work and school and as result frequently choose
to enroll part-time—a known risk factor noted by the U.S. Department of
Education for dropping out (Fry, 2002; Ganderton & Santos, 1995; Hagy &
Staniec, 2002). Moreover, research on the effect of working while in school in
516 The Review of Higher Education Summer 2015

general provides evidence that working more hours can lead to decreases in
grades (Ehrenberg & Sherman, 1987; Stinebrickner & Stinebrickner, 2003),
and may also extend time to program completion (Darolia, 2014).
Overall, the choices undocumented immigrants make for each of these
intensive margin decisions related to when and where to enroll and how to
finance their education are strongly shaped by the low economic resources
of undocumented (and Latino) immigrant youth. Undocumented Latino
youth are likely to delay entry into college, enroll in a 2-year institution, and
work long hours (thus limiting their ability to enroll full-time) because they
do not have the financial resources to pay for the high costs of college—es-
pecially in states that charge out of state tuition. IRT policies should reduce
some of this financial burden and increase the likelihood that undocumented
immigrant students can succeed in college. IRT policies, however, may not
be sufficient given the limited information and lack of financial aid options
available to undocumented immigrant youth.

Research Design
Our contribution to the literature is to analyze the effect of IRT policies
on outcomes beyond enrollment by providing some of the first evidence for
how IRT policies shape the “when,” “where,” and “how” educational behaviors
at a national level. In terms of “when,” we assess the effect IRT policies have
on the enrollment patterns of different age groups and the timing of enroll-
ment (e.g., delayed entry). For “where,” we examine whether IRT policies are
more likely to encourage 2-year vs. 4-year college enrollment. For “how,” we
are interested in how students attend (part-time versus full-time) and how
they finance their education (work behavior and private loans).
Identification Strategy
Our primary empirical approach is to identify the effect of IRT policies on
undocumented student outcomes by comparing the differences in outcomes
of Latino FBNCs (a proxy for undocumented status) covered by the policy to
those not covered, while accounting for outcomes of their U.S. citizen peers
and controlling for variation across states, time, and within states over time.
Consider first the following framework to estimate individual level outcome y:
γits = α + βIts + ηXits + Ss + Tt + λst + eits (1)

Here, i indexes student, t indexes year, s indexes state, α is the intercept, and
e is the error term. I is a binary policy indicator equal to one for state s that
offers in-state tuition to undocumented immigrants in year t. Vectors of
covariates, X, are included, as well as dummy variables for states, S, to con-
trol for time invariant state characteristics (e.g., state-specific educational
Darolia & Potochnick / Educational Implications of IRT Policies 517

policies or stagnant demographic composition) and years, T, to control for

any national time trends that may affect both policy and non-policy states,
such as nationwide changes in college enrollment and national educational
policies (e.g., changes in Pell Grant maximums). State-by-year effects are
captured through a state-year linear trend, , to control for effects specific to
states that are associated with policy passage.
Estimation of equation (1) on the group of Latino FBNC individuals would
allow us to measure the relative outcomes of undocumented students who
have access to in-state tuition compared to those that do not based on the
exogenous variation created by each state’s IRT policy adoption, controlling
for stagnant differences across states, national trends over time, and linear
trends across states over time. However, since we are primarily interested in
recovering estimates of the IRT policy effect on undocumented immigrants,
interpretation of results based on equation (1) does not allow us to rule out
that other unobserved factors (e.g., state-specific shocks or policy changes)
generated outcomes observed by all students in the policy adopting states.
Therefore, we estimate an equation using a pooled sample of Latino FBNCs
and U.S. citizen peers:
γits = α + β1Its + β2Ui + δ(ItsUi ) + ηXits + Ss + Tt + λst + eits (2)

We add the following to equation (1): U, which is a binary indicator equal to

one if the student is a Latino foreign born non-citizen, and the interaction of
U and I. We estimate linear probability models for binary outcome variables
(e.g., enrolled or not) for ease of interpretation and calculate robust standard
errors clustered by state-year for all estimates.4 We consider outcomes start-
ing one year after the policy is enacted to account for the time delay between
when institutions implement the change and students react (Abrego, 2008;
Flores, 2010a, 2010b; Kaushal, 2008).
Our primary variable of interest is the estimated parameter δ, which
is akin to a difference-in-differences-in-differences estimate of the causal
effect of the policy on undocumented immigrants, conditional on covari-
ates. The policy effect can be interpreted as the effect of the IRT policy on
undocumented immigrants covered by the IRT policy. In our identification
strategy, we account for the outcomes of those not directly affected by the
policy (citizens and undocumented immigrants not covered by the policy),
and control for unobserved confounding factors across states, over time, and
within states over time.
We present results in tables from separate estimates comparing the out-
comes of Latino FBNCs against two comparison groups, Latino citizens

4
Results are robust to the use of logit models and are available upon request. The model
is fit by ordinary least squares for continuous variables (e.g., private loan amount).
518 The Review of Higher Education Summer 2015

and all citizens (regardless of race/ethnicity). The use of these comparison

groups allows us to isolate the IRT policy effect by accounting for the parallel
educational trends of citizen students. We select the two comparison groups,
each of which has its own strengths and limitations. Since they are U.S. citi-
zens, neither the Latino citizen nor all citizen comparison groups should be
affected by the IRT law. As members of the same state of residence, however,
both groups should experience similar educational policies and economic
conditions as those experienced by Latino FBNCs. An added strength of the
Latino citizen comparison is that this population—about 60% of whom have
an immigrant parent (Fry & Passel, 2009)—is likely to experience unique
trends associated with race/ethnicity and immigrant status. A limitation of
the Latino citizen comparison, however, is that this group may be affected
by spillover effects. Many Latino citizens live in mixed status families (i.e.,
have undocumented siblings or parents) and thus may benefit from the IRT
policy indirectly since it alters the family budget (Kaushal, 2008). Moreover,
the Latino citizen comparison group may also include undocumented youth
who fear identifying themselves as non-citizens. Given this measurement
error and the potential spillover effects, the Latino citizen comparison may
under-estimate the effect of IRT policies. Thus, we also include the all citizens
comparison group, which should not be affected by spillover effects. In com-
bination, the Latino citizen and all citizen comparison groups should provide
a credible estimate for differencing out state-specific educational trends.
We display results from two different models, and draw primary con-
clusions from results that are robust across specifications. The first model
includes just the indicators for having an IRT policy and the Latino undocu-
mented student proxy, along with the interaction between them, and state
and year controls. In the second, we add the state-year linear trend and a
vector of covariates, X, with parameter vector, η. We include individual-level
controls for age, gender, race/ethnicity, and marital status, and controls for
state-level time varying characteristics that may be correlated with policy
adoption, such as macroeconomic conditions measured by the unemploy-
ment rate, and educational trends (educational attainment levels of whites
and Latinos).5

5
We strive to create similar models for comparability, but there are some differences between
the models using CPS and NPSAS data because of data availability. Monthly unemploy-
ment rate is included in the CPS data analysis, while yearly unemployment rate is included
in the NPSAS data analysis. In the models using CPS data, we also include month controls
to account for variation in college enrollment across months (e.g., the lower likelihood of
enrolling or graduating during the summer months) and the number of years individuals
have been in the U.S.
Darolia & Potochnick / Educational Implications of IRT Policies 519

Data
We use two large nationally representative data sources in the paper. Similar
with prior research on the effects of IRT policies on college behavior (Flores,
2010a, 2010b; Kaushal, 2008), we use the Merged Outgoing Rotation Group
(MORG) file from the Current Population Survey (CPS), a nationally rep-
resentative sample sponsored by the U.S. Census Bureau and U.S. Bureau of
Labor Statistics. Using a multistage stratified sample, the CPS collects monthly
demographic, employment, and enrollment information from about 60,000
housing units across the United States for the civilian population age sixteen
and older. We focus on a sub-sample of high school completers, because to
qualify for an IRT policy, an individual for most states has to complete high
school in that state.
The second data source, the National Postsecondary Student Aid Study
(NPSAS), is a nationally representative cross-section of college students who
attend Title IV eligible postsecondary institutions. We use data on under-
graduate students from four NPSAS waves to account for the time period
when a number of states considered IRT policies, 2000, 2004, 2008, and 2012.
NPSAS data include between approximately 60,000–114,000 records for
each of these waves. In these data, we observe each student’s demographic
and enrollment characteristics, as well as measures of financial need and
methods of college payment.
For both datasets we rely on a proxy to identify undocumented im-
migrants, Latino foreign-born, non-citizens (FBNC). Because no national
research survey collects information on documentation status (Passel, 2005),
researchers have relied on proxies for undocumented status (Bozick & Miller,
2014; Chin & Juhn, 2011; Dickson & Pender, 2013; Flores, 2010a; 2010b;
Kaushal, 2008; Potochnick, 2014 ). Given that about 80% of undocumented
immigrants are Latino (Passel & Cohn, 2008), FBNC Latino is one of the
strongest proxies available.6 Nevertheless, a limitation of using a proxy to
identify undocumented status is that our estimates are likely to be down-
wardly biased since our estimate of the policy effect includes individuals
unlikely to be affected by the policy (i.e., LPRs). Moreover, though both data
sets are large in total observations, the relatively smaller samples of Latino
FBNCs (~12,000 in CPS and ~5,500 in NPSAS), may limit our power when
trying to precisely identify policy effects. Because measurement error and
small sample sizes increase the variance of the estimate and the probability
of a Type II error—failing to reject the null hypothesis of no effect when the
policy actually has an effect (Wooldridge 2010)—we follow the convention

6
For CPS, we strengthen this proxy by excluding individuals who migrated to the U.S.
before 1986 since these individuals likely received an adjustment of status under the 1986
federal immigration reform.
520 The Review of Higher Education Summer 2015

of prior research on IRT policies and report p-values at the .10 level (Flores
2010a; Kaushal 2008).
There are several important differences between the CPS and NPSAS
datasets that may contribute to variation in results across the datasets. Given
their different sampling frames, for instance, the two datasets differ in their
likelihood of including labor migrants (i.e., individuals who come to the U.S.
to work and never enter the school system; Oropesa & Landale, 2009) who
are less likely to respond to educational policies. Because the NPSAS sample
is based on college enrollees the labor migrant sample is likely to be less of a
confounder than the sample in CPS, which may explain some of the variation
in results across the datasets. Additionally, differences in survey frequency
(monthly vs. every 4 years) and survey focus (labor vs. postsecondary trends)
may also contribute to variation in results across the datasets.
In Table 2, we display summary statistics for Latino FBNCs, Latino citizens,
and all citizens from NPSAS and CPS. We list summary statistics for the CPS
sample for outcomes and controls used in the subsequent analyses, i.e., from
the 18–24 year old sample we display summary statistics for enrollment and
financing, and from the 26–28 year old sample we display averages related
to the attainment analysis. There are key similarities and differences among
Latino FBNCs, Latino citizens, and all citizens. In both age groups of CPS,
Latino FBNCs are more likely to be male and married than Latino citizens
and all citizens. Among 18–24 year olds, Latino FBNCs are less likely to be
enrolled in college (21%) than Latino citizens (41%) and all citizens (40%)—
particularly full-time (15% vs. 35% vs. 34%, respectively). In the NPSAS
sample that includes college enrollees only, Latino FBNCs are actually more
likely to be enrolled full-time (50%) than their Latino citizen (46%) and all
citizen (42%) counterparts, but also more likely to be enrolled in a 2-year
college (65% vs. 59% vs. 49%, respectively). In terms of work, we observe in
both NPSAS and CPS that Latino FBNCs have similar employment levels
compared to their citizen peers but that Latino FBNCs work more hours
on average. Lastly, among the older CPS sample aged 26–28, we unsurpris-
ingly see that academic attainment is lower among Latino FBNCs than their
citizen counterparts.

Findings
When to Attend
We begin with our findings related to whether the policies affect when
students attend college in Table 3. In all tables, we display the coefficient
on our policy effect variable (Latino FBNC X post-policy), which can be
interpreted as the IRT policy effect on Latino FBNCs. Full regression output
is available upon request, with coefficients on control variables generally
Darolia & Potochnick / Educational Implications of IRT Policies 521

Table 2
Weighted Sample Summary Characteristics
Latino FBNC Latino Citizens All Citizens
Mean SD Mean SD Mean SD

CPS Enrollment Sample, Age 18–24

Age 21.51 (1.87) 21.24 (1.90) 21.23 (1.90)
Female 0.45 (0.50) 0.51 (0.50) 0.51 (0.50)
Average years in U.S. 7.23 (5.30) 21.08 (2.29) 21.03 (2.39)
Married 0.25 (0.44) 0.11 (0.31) 0.11 (0.32)
Enrolled in college 0.21 (0.40) 0.41 (0.49) 0.40 (0.49)
Enrolled part-time 0.05 (0.23) 0.06 (0.23) 0.06 (0.23)
Enrolled full-time 0.15 (0.36) 0.35 (0.48) 0.34 (0.48)
Employed 0.65 (0.48) 0.66 (0.47) 0.65 (0.48)
Hours worked1 37.29 (8.84) 33.71 (11.59) 33.73 (11.48)
N= 11,428 389,776

CPS Attainment Sample, Age 26–28

Age 27.00 (0.82) 27.00 (0.82) 27.00 (0.82)
Female 0.44 (0.50) 0.51 (0.50) 0.51 (0.50)
Average years in U.S. 7.28 (4.80) 26.63 (2.51) 26.53 (2.78)
Married 0.55 (0.50) 0.42 (0.49) 0.42 (0.49)
Obtained Associate’s 0.03 (.17) 0.05 (.23) 0.05 (0.23)
degree
Obtained Bachelor’s 0.13 (.34) 0.36 (.48) 0.34 (0.47)
degree
N= 12,797 164,765 358,426

NPSAS Sample
Age 25.70 (8.00) 25.30 (8.36) 26.37 (9.66)
Female 0.59 (0.49) 0.59 (0.49) 0.57 (0.49)
Married 0.26 (0.44) 0.20 (0.40) 0.22 (0.41)
Delayed Entry Years 2.59 (4.99) 1.80 (4.44) 2.05 (5.20)
Enrolled part-time 0.50 (0.50) 0.54 (0.50) 0.58 (0.49)
Enrolled full-time 0.50 (0.50) 0.46 (0.50) 0.42 (0.49)
Enrolled in a ≤2-yr 0.65 (0.48) 0.59 (0.49) 0.49 (0.50)
college
Enrolled in a 4-yr 0.35 (0.48) 0.41 (0.49) 0.51 (0.50)
college
Has a private loana 0.23 (0.42) 0.35 (0.48) 0.39 (0.49)
Private loan 5,821 (5,317) 6,323 (5,981) 6,882 (5,944)
amount
Employed 0.51 (0.50) 0.53 (0.50) 0.53 (0.50)
Hours workedb 31.03 (13.07) 30.45 (13.34) 29.75 (13.92)
N= 5,470 38,740 300,790

Note. Unweighted NPSAS counts are rounded to the nearest 10. Survey weights used for sample char-
acteristics. Source data: Current Population Survey 1998–2012 and the National Postsecondary Student
Aid Study 2000, 2004, 2008, and 2012.
a
Sample is only those with private loans
b
Sample is only those employed.
522 The Review of Higher Education Summer 2015

Table 3
Impact of IRT Policies on “When” Decisions, Latino For-
eign Born Non-Citizens
Compared to Compared to
Latino Citizens All Citizens
(1) (2) (3) (4)

A. College Enrollment by Age Group

Enrolled in College (Age 18–24) 0.0194 0.0151 0.0235* 0.0217*
(0.0121) (0.0115) (0.0135) (0.0113)
Enrolled in College (Age 18–20) 0.0554** 0.0357* 0.0608*** 0.0499***
(0.0207) (0.0199) (0.0208) (0.0187)
Enrolled in College (Age 21–24) 0.0040 0.0023 0.0022 0.0059
(0.0116) (0.0118) (0.0125) (0.0117)

B. Delayed Entry
Years Post-HS Before -0.7905*** -0.6090*** -0.7464*** -0.5923***
Entering College (0.2490) (0.1841) (0.2694) (0.2142)

State and Year Indicators X X X X

Additional Controls X X

Note. (1) Additional demographic and state controls include: age, female, marital status, race/ethnicity,
unemployment rate, the proportion of non-Hispanic white adults with some college in the state, and the
proportion of Hispanic adults with a high school diploma in the state, as well as state and year indicators
and a state-year linear trend; The CPS sample also includes controls for month and average years in U.S.;
(2) Standard errors are adjusted for clustering by state-year; (3) survey weights used; (4) Unweighted
NPSAS counts are rounded to the nearest 10. Source data: Panel A: Current Population Survey 1998–2012
[Age 18–24 (N=50,456 in Latino FBNC & Latino citizen sample and 401,204 in Latino FBNC & all citizen
sample), Age 18-20 (N=19,930 and 156,818), Age 21-24 (N=30,526 and 244,386)]; Panel B: the National
Postsecondary Student Aid Study 2000, 2004, 2008, and 2012 waves (N=44,210 and 306,260).
* p<.10, ** p<.05, ***p<.01.

indicating expected relationships. The first two columns include results from
the two models using a sample of Latino FBNCs and Latino citizens, while
the third and fourth columns display results from the same models using a
sample of Latino FBNCs and all citizens.
For comparability against prior research, we first present estimates of the
effect of IRT policies on college enrollment among students who are 18–24
years old. Our results are consistent with studies that find IRT policies in-
crease the likelihood of college enrollment (Flores, 2010a, 2010b; Kaushal,
2008). For 18–24 year olds, we find point estimates of about two percentage
points in models including both Latino and all citizens, though the effect is
only statistically significant at the 90% confidence level (CL) when compared
to the latter group. This increase appears to be driven by the enrollment of
younger students, as we observe a larger effect when examining 18–20 year
Darolia & Potochnick / Educational Implications of IRT Policies 523

olds, reaching four to six percentage points. A four to six percentage point
effect indicates that the policies increase enrollment by about 19–29% off
the Latino FBNC sample enrollment rate of about 21%. When examining
older students (21–24 year olds), we do not detect a policy effect. This sug-
gests, therefore, that IRT policies have a larger influence on the enrollment
behavior of younger students.
Results in Panel B suggest one mechanism that leads to the younger student
enrollment increase from Panel A. Here, we estimate IRT policy effects on
the number of years students delay entry into college after completing high
school. Though unsurprisingly Latino FBNC students have a longer average
duration of delayed entry into college than their citizen peers (results not
shown), we find that Latino FBNC students reduce the number of years they
delay entry into college by about half to three-quarters of a year. All of these
coefficients are statistically significant at the 99% CL. Therefore, IRT poli-
cies appear to encourage students to enter college earlier. This earlier entry
could be because the lower costs of college allow students to enroll without
first accumulating substantial savings. Another motivating factor could be
changes to the higher education, school, or community contexts because of
IRT policies. In response to the IRT policies, these communities may seek
to provide more information to undocumented youth about how to attend
and afford college—key factors in the college decision process (Perna, 2006)
Where to Attend
We next consider whether IRT policies affect decisions about where to
attend college. In Table 4, we display IRT policy effects on students’ enroll-
ment and credential receipt from 4-year or 2-year colleges. Negative point
estimates provide weak directional evidence that students may be less likely
to attend a 4-year college, as opposed to a 2-year or less college, but these
effects are not precisely estimated.
We examine the 2-year versus 4-year college decision further by analyzing
the effect of IRT policies on type of credential students obtained.6 We find
negative point estimates for an IRT effect on earning a bachelor’s degree of
almost 16 percentage points (as compared to earning an associate’s degree,
for those that obtain at least an associate’s degree), when compared to all
citizens in the third and fourth columns. This indicates that IRT policies
are encouraging completion of an associate’s degree at a higher rate than a
bachelor’s degree. This trend is consistent with the directional influence we
found in our 2-year vs. 4-year assessment as well as descriptive research that

7
We examine outcomes related to program completion among 26–28 year olds. Because
a large share of individuals in this age group are likely to be labor migrants, we exclude im-
migrants who arrived to the U.S. after age 15 (Oropesa & Landale, 2009).
524 The Review of Higher Education Summer 2015

Table 4
Impact of IRT Policies on “Where” Decisions, Latino
Foreign-Born Non-Citizens
Compared to Compared to
Latino Citizens All Citizens
(1) (2) (3) (4)

Enrolled in a 4-yr College vs. 0.0051 -0.0018 -0.0286 -0.0197

2-yr College
(0.0248) (0.0230) (0.0358) (0.0329)
Obtained Bachelor vs. -0.0820 -0.0753 -0.155** -0.1587**
Associate Degree
(0.0714) (0.0712) (0.0716) (0.0707)
State and Year Indicators X X X X
Additional Controls X X

Note. (1) Additional demographic and state controls include: age, female, marital status, race/ethnicity,
unemployment rate, the proportion of non-Hispanic white adults with some college in the state, and the
proportion of Hispanic adults with a high school diploma in the state, as well month indicators and state
and year linear trends; (2) Standard errors are adjusted for clustering by state-year; (3) survey weights
used, (4) CPS sample excludes Latino FBNCs who arrived after age 15. Source data: 4-year vs. 2-year
enrollment: the National Postsecondary Student Aid Study 2000, 2004, 2008, and 2012 waves (N=44,210
in Latino FBNC & Latino citizen sample and 306,260 in Latino FBNC & all citizen sample); Bachelor vs.
associate degree: Current Population Survey 1998-2012, among those that obtained an associate degree
or higher, ages 26-28 (N = 5,237 and N = 82,146).
* p<.10, ** p<.05, ***p<.01.

suggests community colleges are the choice institution of IRT students and
Latinos in general (Flores, 2010b; Teranishi et al., 2011). This result also aligns
with Kaushal’s (2008) assessment, which found that IRT policies have had
a small impact on the likelihood of attaining an associate’s degree, though
she did not assess the impact on bachelor degree attainment. Overall, our
results suggest that while IRT policies may increase educational enrollment,
undocumented students still face barriers to completing a degree, especially
at a bachelor’s level.
How to Attend and Finance
Table 5 moves to analysis of how students finance college. Since the CPS
and NPSAS data overlapped for some of these outcomes, we display results
from both models where applicable. We begin with an examination of
whether IRT policies affect whether enrolled college-goers attend part-time
or full-time. All point estimates are approximately two to four percentage
points, though results are only statistically significant (at the 90% CL) in one
model, when compared to Latino citizens using the adjusted model with the
NPSAS sample. An interpretation of these estimates is that IRT policies may
Darolia & Potochnick / Educational Implications of IRT Policies 525

Table 5
Impact of IRT Policies on “How” Decisions, Enrolled La-
tino Foreign Born Non-Citizens
Compared to Compared to
Latino Citizens All Citizens
(1) (2) (3) (4)

A. Enrollment Pattern
Full-Time vs. Part-Time (CPS)a -0.0204 -0.0251 -0.0189 -0.0220
(0.0206) (0.0224) (0.0206) (0.0208)
Full-Time vs. Part-Time (NPSAS)b -0.0302 -0.0387* -0.0275 -0.0315
(0.0276) (0.0218) (0.0294) (0.0235)

B. Employment
Employed (CPS)c -0.0200 -0.0196 -0.0193 -0.0261
(0.0253) (0.0255) (0.0241) (0.0243)
Employed (NPSAS)d 0.0323* 0.0324* 0.0340* 0.0314*
(0.0175) (0.0174) (0.0183) (0.0179)
Number of Hours Worked (CPS)e -1.2123 -1.1962 -1.4321* -1.5851*
(0.8312) (0.8029) (0.8556) (0.8119)
Number of Hours Worked (NPSAS)f -0.1756 -0.1321 -0.0218 -0.1480
(0.6881) (0.6171) (0.7452) (0.6138)

C. Private Loan Borrowing

Has a Private Loand 0.0030 0.0036 0.0067 0.0082
(0.0089) (0.0083) (0.0088) (0.0081)
Private Loan Amountg 884.5566 361.1509 799.1285 381.2156
(879.7108) (859.0980) (742.4968) (685.5479)

State and Year Indicators X X X X

Additional Controls X X

Note. (1) Additional demographic and state controls include: age, female, marital status, race/ethnicity,
unemployment rate, the proportion of non-Hispanic white adults with some college in the state, and the
proportion of Hispanic adults with a high school diploma in the state, as well as state and year indicators
and a state-year linear trend; The CPS sample also includes controls for month and average years in U.S.;
(2) Standard errors are adjusted for clustering by state-year; (3) survey weights used; (4) Unweighted
NPSAS counts are rounded to the nearest 10. Source data: Current Population Survey 1998-2012 and
the National Postsecondary Student Aid Study 2000, 2004, 2008, and 2012.
a
CPS sample (N=17,048 in Latino FBNC & Latino citizen sample and 158,954 in Latino FBNC & all
citizen sample). b NPSAS sample ((N=44,210 and 306,260). c CPS sample, (N=18,160 and 164,658).
d
NPSAS sample (N=44,210 and 306,260). e If worked; CPS sample (N= 8,230 and 77,006). f If worked;
NPSAS sample (N = 19,690 and 136,760). g If borrowed private loan money; NPSAS sample (N = 3,290
and 28,810).* p<.10, ** p<.05, ***p<.01.
526 The Review of Higher Education Summer 2015

induce new part-time enrollment at a mildly higher rate than it encourages

students to switch from part-time to full-time enrollment.
We next present results from models that examine how IRT policies affect
how much students work and borrow. A concern is that financially con-
strained undocumented students will be induced to attend college because
of the lower tuition associated with IRT policies, but these students may still
have difficulty meeting all educational and non-educational costs accrued
while attending college. Taken together, we find only suggestive evidence
that enrolled Latino FBNCs are working or borrowing more post-policy.
The lack of conclusive results may be because of difficulties undocumented
immigrants confront in obtaining financial resources to pay for college ex-
penses from external sources, such as employment or credit.
We find mixed evidence that IRT policies affect working behavior. We find
no precisely estimated results using CPS, though all point estimates are nega-
tive. Results from NPSAS indicate that students are about three percentage
points more likely to work after the passage of an IRT policy, though this
result is statistically significant at only the 90% CL. This may be because the
policies encourage working students to attend college, which corresponds
with an increase in the proportion of part-time students. Both the CPS and
NPSAS data samples point to fewer hours worked because of IRT policies,
with the results using CPS data larger in magnitude and only the results
comparing Latino FBNC students to all citizens are statistically significant
(at the 90% CL, IRT policy effect of about 1.5 hours worked fewer in Panel
C, columns 3 and 4). As shown at the bottom of Panel C, results point to
students possibly being more likely to borrow private student loan money
and of larger magnitudes, but these results are not statistically significant.
Alternative Non-Policy State Comparison
We also run models where we include only states that considered an IRT
policy but did not pass the policy (see Potochnick, 2014 for a full list) in the
comparison group, since these states may be more similar to IRT policy states
that states that did not consider the policy. The results presented in Table 6
indicate that our results are robust to the different state comparison group.
The point estimates for all outcomes are similar (sometimes slightly larger
and sometimes slightly smaller) to those of our original estimates for both
comparison groups and both datasets. For example, when examining col-
lege enrollment in the CPS sample (Panel A), we still find point estimates of
about four to six percentage points in the model including all citizens, and the
result remains significant. The results for delayed entry continue to indicate
a policy effect around 5.5 percentage points or greater across all comparison
groups. The point estimate for full-time enrollment in models including
Latino citizens using NPSAS data increased but is no longer significant.
Darolia & Potochnick / Educational Implications of IRT Policies 527

The estimated change in probability of being employed increases by al-

most a percentage point using the NPSAS data, while the change in number
of hours worked in the CPS sample is no longer significant at conventional
levels (the point estimates for all citizens (column 4) remains comparable
to our previous estimate). In terms of degree obtainment, our new point
estimates are actually larger and more precisely estimated than our original
estimates when comparing undocumented immigrants to Latino citizens –
nearly a six percentage point increase in probability of obtaining at least an
associate’s degree with the alternative state comparison group.
Overall, the results of this robustness check and the general consistency
of our estimates across two different national samples (CPS and NPSAS)
and three different comparison groups (Latino FBNCs in policy vs. non-
policy states; Latino FBNCs vs. Latino citizens; Latino FBNCs vs. all citizens)
provides strong evidence for the policy effects detected. These policy effects
remained robust to the inclusion of state-specific linear trends and were
consistent when using logit regression (for dichotomous outcomes) rather
than a linear probability model (results available upon request). We also
did not find notable differences among states that are new or traditional
immigrant destinations.8
Limitations
Nonetheless, our paper faces several limitations, many which were dis-
cussed throughout the text. We summarize the primary limitations here. First,
because we rely on Latino FBNCs as a proxy for undocumented status the
sample includes individuals not in the treatment group. Consequently, the
policy effect estimates are likely to be attenuated. Second, each of the data-
sets used in this analysis have key limitations. NPSAS only include students
enrolled in college, so results based on these data should be considered as the
effect of the IRT policy conditional on enrollment. While CPS is not condi-
tional on enrollment, it lacks detailed contextual information about other
college choices. Lastly, if policy states adopted other policies in addition to
the IRT policy to address educational needs of undocumented immigrants,
our policy effect would be biased and confound the IRT policy effect with

8
Because the availability of information and support systems for the IRT policies may be
greater in states with a longer migration history, we also ran our analysis separately for new
and traditional immigrant destination states based on the work of Flores and Chapas (2009).
Using their same classification of new and traditional states we did not find consistent evidence
that there were differential effects between new and traditional states. In only a few instances
(enrollment and employment) we found stronger effects for FBNCs in new destination states
compared to traditional states. This analysis is available upon request. Because a larger share
of the immigrant population in new destination states is undocumented (Passel & Cohn,
2008), these larger effects likely reflect the fact that our FBNC proxy is stronger in these states.
528 The Review of Higher Education Summer 2015

Table 6
IRT Policies Impact on Latino FBNCs, Using Only States
that Considered the Policy
Compared to Compared to
Latino Citizens All Citizens
(1) (2) (3) (4)

Enrolled in College (CPS)

Age 18–24 0.0166 0.0118 0.0229 0.0199*
(0.0124) (0.0118) (0.0139) (0.0117)
Age 18–20 0.0502* 0.0301 0.0601*** 0.0472**
(0.0212) (0.0206) (0.0213) (0.0192)
Age 21–24 0.0032 0.0007 0.0012 0.0042
(0.0121) (0.0123) (0.0131) (0.0122)
Delayed Entry (NPSAS) -0.8299*** -0.5792** -0.7690** -0.5555*
(0.3029) (0.2493) (0.3388) (0.2918)
Enrolled Full-Time vs. Part-Time -0.0262 -0.0302 -0.0207 -0.0243
(CPS) (0.0211) (0.0231) (0.0211) (0.0214)
Enrolled Full-Time vs. Part-Time -0.0326 -0.0441 -0.0320 -0.0329
(NPSAS) (0.0410) (0.0303) (0.0427) (0.0316)
Enrolled in a 4-yr College (NPSAS) 0.0045 0.0101 -0.0389 -0.0169
(0.0300) (0.0306) (0.0474) (0.0473)
Employed (CPS) -0.0214 -0.0221 -0.0191 -0.0255
(0.0260) (0.0264) (0.0249) (0.0250)
Employed (NPSAS) 0.0421* 0.0402* 0.0484** 0.0430*
(0.0235) (0.0234) (0.0226) (0.0231)
Number of Hours Worked -0.8036 -0.8473 -1.0431 -1.2426
(if worked, CPS) (0.8428) (0.8131) (0.8754) (0.8310)
Number of Hours Worked -0.4984 -0.2984 -0.0287 -0.1639
(if worked, NPSAS) (0.7954) (0.6975) (0.8903) (0.6943)
Has a Private Loan (NPSAS) 0.0039 0.0038 0.0073 0.0080
(0.0107) (0.0095) (0.0106) (0.0093)
Private Loan Amount 970.8093 607.4896 915.0578 586.2558
(if has loan, NPSAS) (906.6271) (718.5916) (791.7054) (647.8818)
Bachelor’s vs. Associate’s Degree
(if obtained degree, CPS) -0.0582 -0.0537 -0.1246 -.1261*
(0.0748) (0.0753) (0.0756) (0.0745)
State and Year Indicators X X X X
Additional Controls X X

Note. (1) 20 States have considered IRT legislation. See Potochnick (2014) for a list of states. (2) Additional
demographic and state controls include: age, female, marital status, race/ethnicity, unemployment rate, the
proportion of non-Hispanic white adults with some college in the state, and the proportion of Hispanic
adults with a high school diploma in the state, as well as state and year indicators and a state-year linear
trend; The CPS sample also includes controls for month and average years in U.S.; (3) Standard errors are
adjusted for clustering by state-year; (4) survey weights used; (5) Unweighted NPSAS counts are rounded
to the nearest 10. Source data: Current Population Survey 1998-2012 and the National Postsecondary
Student Aid Study 2000, 2004, 2008, and 2012.
* p<.10, ** p<.05, ***p<.01.
Darolia & Potochnick / Educational Implications of IRT Policies 529

the effect of these other policies. Research on policy adoption as well as the
falsification checks included in this study limit the plausibility of this threat
(Reich & Barth 2010; Reich & Mendoza 2008).

Discussion
The labor market preparation of a state’s undocumented citizenry is
becoming an increasingly important issue given that federal action has al-
ready begun to reduce legal barriers for undocumented youth. Beginning in
2012, undocumented youth enrolled in school or who have received a high
school diploma/GED certificate can apply for deportation waivers and work
permits under the Deferred Action for Childhood Arrivals (DACA) execu-
tive order. Efforts to extend deportation protection to other undocumented
immigrants via executive action are on-going (Shear, 2014). Additionally, the
federal government continues to debate national immigration reforms that
would provide a pathway to citizenship, particularly for college educated
undocumented youth under the Federal DREAM Act.9 As the federal land-
scape continues to change, states policymakers could benefit from a more
comprehensive understanding of the effects of state adopted in-state resident
tuition policies that reduce financial constraints for undocumented students.
This study builds on prior research that examines the effect of IRT policies
on undocumented immigrant’s college investment decision by providing the
first in-depth assessment of when and where students enroll, and how they at-
tend and finance their education. Given the financial obstacles and uncertain
labor prospects that undocumented immigrant youth face, IRT policies may
not produce the same private and public benefits typically associated with
public subsidies for higher education. Thus, to assess whether IRT policies
are likely to achieve optimal returns for the states that have adopted them
we examine key indicators of the educational investment decision.
Taken together, our results, which are based on two large national data-
sets, suggest that while IRT policies increase enrollment of undocumented
immigrants in postsecondary education, further consideration is needed

9
The Development, Relief, and Education for Alien Minors Act (known as the DREAM
Act) has been under consideration in congress since 2001. If passed, it would enact two
major changes to current federal law: 1) provide a pathway to citizenship for students who
came to the U.S. at or before age 15 and who met certain criteria, and 2) eliminate the federal
provision that penalizes states for providing in-state tuition without regard to immigration
status (National Immigration Law Center, 2009). However, if adopted, the DREAM Act will
not resolve the debate over whether undocumented immigrants should qualify for in-state
tuition, because the bill does not require states to provide in-state tuition to undocumented
immigrants.
530 The Review of Higher Education Summer 2015

beyond just whether more students attend. Our study corroborates prior
studies (Flores, 2010a, 2010b; Kaushal, 2008) that IRT policies lead to in-
creased enrollment among undocumented immigrant students but also finds
that the policies affect the decision of when to enroll. We observe that IRT
policies encourage students to enroll in college sooner after college, with the
time between high school completion and college enrollment reduced by
about half to three quarters of a year. Providing further evidence that IRT
policies reduce delayed entry, we find that increases in enrollment related
to IRT policies are driven by the enrollment of younger students. This can
be an important improvement for many students. Extant research suggests
that reductions in the amount of time undocumented students wait to enter
college should increase their chances for college success (Ganderton & Santos,
1995; Stratton et al., 2008). Moreover, going to college sooner allows students
to begin to earn the higher college degree wages sooner. This provides both
immediate benefits because of augmented current income, but also benefits
over workers’ lifetimes due to the compounding nature of earnings increases.
In terms of where students are attending, our estimates provide only
weak evidence that IRT policies have a stronger effect on enrollment in the
2-year vs. 4-year college sector. We find, however, that increased enrollment
is primarily leading to associate’s, rather than bachelor’s, degree attainment.
This is consistent with previous findings that Latinos in general are more
likely to enroll in community colleges since these colleges are more afford-
able and accessible than 4-year colleges (Teranishi et al., 2011). Lowering
the direct costs of 4-year colleges, however, may also lead new enrollees and
2-year college students to attend or transfer to 4-year programs. Research
in Texas, for instance, suggests that while the states IRT policy has had a
particularly strong influence on community college enrollment it has also
increased attendance at more selective 4-year institutions (Dickson & Pender,
2013; Flores, 2010b). Thus, the evidence to date suggests that admission and
financial aid policies affecting undocumented youth are relevant at both
2-year and 4-year institutions.
Our assessment of how students are financing their education indicates
some potential limitations of IRT policies. For example, we find suggestive
evidence that IRT policies may encourage non-enrollees to enroll part-time
at a higher rate than it induces part-time enrollees to move to full-time study.
This may be because the lower tuition associated with IRT policies may not
be a sufficient financial benefit to allow many students to overcome the work
and family obligations or resource constraints that prevent them from being
able to attend full-time.
Our analysis of working and borrowing behavior provides further infer-
ence on this. In particular, if IRT policies induce financially constrained stu-
dents into college, then the financial needs of this new student group need to
Darolia & Potochnick / Educational Implications of IRT Policies 531

be considered as they may face unique challenges among students. We observe

suggestive post-policy evidence from the NPSAS data that students are more
likely to work while enrolled. The effect was not as strong as we expected,
perhaps because undocumented students also face legal challenges to obtain-
ing employment. We do not find evidence of changes in borrowing behavior
because of IRT policies, which may in part be due to limited credit availability
for the undocumented. Nevertheless, working behavior and borrowing needs
need to be carefully monitored among undocumented students. Research on
the effect of working while in school provide evidence that working more
hours can potentially have adverse consequences to academic performance
(Darolia, 2014; Ehrenberg & Sherman, 1987). Moreover, borrowing might
be particularly burdensome on undocumented immigrants since they face
uncertain job prospects due to their legal status (Suarez-Orozco et al., 2011)
and are expected to make less than their authorized peers (Hall, Greenman,
& Farkas, 2010).
Overall, our results give rise to concern that the adoption of an IRT policy
may not be sufficient to ensure undocumented immigrant’s success in higher
education. Undocumented immigrants who are responding to the IRT policy
and enrolling in college appear to be making both optimal and sub-optimal
investment decisions that will affect their chances of college completion.
On the positive side, undocumented immigrants are attending college at a
younger age. On the negative side, they are enrolling part-time in less selec-
tive 2-year institutions and face competing work demands. These intensive
margin educational investment decisions are likely shaped by the economic
and legal constraints (Greenman & Hall, 2013) and limited informational
knowledge (Teranishi et al., 2011) undocumented immigrants continue to
face even in the wake of an IRT Policy. IRT policies only address part of the
college choice by lowering tuition rates but do not address other contextual
factors (e.g., access to knowledge, poverty, cultural preferences, and the
broader policy context) that guide college choices (Perna, 2006; Portes &
Rumbaut, 2006).
Addressing these other contextual factors could be beneficial for states
as well. There is evidence of returns to going to college without comple-
tion, though completing degrees provides students with the most potential
benefits from college attendance (Belfield & Bailey, 2011). Amplified enroll-
ment comes at a cost to the state, including increased outlays to support the
educational endeavors of publicly subsidized students. Therefore, if states
that adopt an IRT policy want to fully maximize returns on their investment,
our results suggest the need for further financial and academic assistance
to help students finish associate’s degrees and encourage bachelor’s degree
completion. One potential avenue is to extend access to state financial aid
and private scholarships—a policy solution that has only been adopted by
a few states but rapidly gaining traction in others.
532 The Review of Higher Education Summer 2015

Furthermore, to reap the full returns to their investment in IRT policies,

policymakers should consider undocumented immigrants’ employment op-
tions. In particular, these highly skilled workers need to have incentives to
reside in-state post-college. Consequently, in conjunction with educational
policies, states should consider policies that allow for these productive un-
documented immigrant students to formally participate in labor markets and
ease other barriers (such as allowing undocumented individuals to obtain
driver’s licenses so they can travel to and from work). When students leave
the state after increasing their skills through publicly funded education, this
leakage of enhanced productivity can diminish benefits associated with states’
funding. Thus, as states continue to debate college access for undocumented
students, our results suggest that they would benefit from taking a more
comprehensive approach that addresses financial barriers beyond in-state
tuition as well as future employment limitations that shape undocumented
immigrants’ educational investment decision.

References
Abrego, L. (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. (2008). Legitimacy, social identity, and the mobilization of law: The ef-
fects of Assembly Bill 540 on undocumented students in California. Law &
Social Inquiry, 33(3), 709–734.
Avery, C. & Turner, S. (2012). Student loans: Do college students borrow too much-
or not enough? Journal of Economic Perspectives, 26(1), 165–192.
Becker, G. S. (1964). Human capital: A theoretical and empirical analysis. New York:
Columbia University Press.
Belfield, C. R. & Bailey, T. (2011). The benefits of attending community college: A
review of the evidence. Community College Review, 39(1), 46–68.
Ben-Porath, Y. (1967). The production of human capital and the life cycle of earn-
ings. The Journal of Political Economy, 75(4), 352–365.
Bozick, R., & Miller, T. (2014). In-state college tuition policies for undocumented
immigrants: Implications for high school enrollment among non-citizen
Mexican youth. Population Research and Policy Review, 33, 13–30.
Chin, A., & Juhn, C. H. (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 universi-
ties. In D. L. Leal & S.J. Trejo (Eds.), Latinos and the economy, integration and
impact in schools, labor markets, and beyond (pp. 63–94). New York: Springer.
Conger, D. (2014). The impact of tuition increases on undocumented college stu-
dents’ enrollment and grades. Unpublished manuscript.
Conger, D. & Chellman, C. C. (2013). Undocumented college students in the United
States: In-state tuition not enough to ensure four-year degree completion.
Education Finance & Policy, 8(3), 364–377.
Darolia & Potochnick / Educational Implications of IRT Policies 533

Contreras, F. (2009). Sin papeles y rompiendo barreras: Latino students and the
challenges of persisting in college. Harvard Educational Review, 79(4), 610–631
Darolia, R. (2014). Working (and studying) day and night: Heterogeneous effects of
working on the academic performance of full-time and part-time students.
Economics of Education Review, 38, 38–50.
Dickson, L. & Pender, M. (2013). Do in-state tuition benefits affect the enrollment
of non-citizens? Evidence from universities in Texas. Economics of Education
Review, 37, 126–137.
Dynarski, S. (2000). Hope for whom? Financial aid for the middle class and its impact
on college attendance. National Tax Journal, 53(3), 629–662.
Dynarski, S. (2004). The new merit aid. In C. M. Hoxby (Ed.), College choices: The
economics of where to go, when to go, and how to pay for it (pp. 63–100). Chicago:
University of Chicago Press.
Ehrenberg, R. & Sherman, D. (1987). Employment while in college, academic achieve-
ment, and postcollege outcomes: A summary of results. Journal of Human
Resources, 22(1), 1–23.
Flores, S. M. (2007). The effect of in-state resident tuition policies on the college enroll-
ment of undocumented Latino students in Texas and the United States. Disserta-
tion, Harvard University.
Flores, S. M. (2010a). State Dream Acts: The effect of in-state resident tuition poli-
cies and undocumented Latino students. Research in Higher Education, 33(2),
239–283.
Flores, S. M. (2010b). The first state Dream Act: Immigration and in-state resident
tuition in Texas. Educational Evaluation and Policy Analysis, 32, 435–55.
Flores, S. M., & Chapas, J. (2009). Latino immigrant access to higher education in a
bipolar context of reception. Journal of Hispanic Higher Education 8(1), 90–109.
Fry, R. (2002). Latinos in higher education: Many enroll, too few graduate. Washington,
DC: Pew Hispanic Center.
Fry, R., & Passel, J. (2009). Latino children: A majority are U.S.-born offspring of im-
migrants. Washington, DC: Pew Hispanic Center.
Ganderton, P. T. & Santos, R. (1995). Hispanic college attendance and completion:
Evidence from the High School and Beyond Surveys. Economics of Education
Review, 14(1), 35–46.
Gonzales, R. (2008). Left out but not shut down: Political activism of undocumented
student movement. Northwestern Journal of Law & Social Policy 3(2), 219–239.
Gonzales, R. (2009). Young lives on hold: The college dreams of undocumented students.
New York: The College Board.
Gonzales, R. (2012). In spite of the odds: Undocumented immigrant youth, school
networks, and college success. In C. G. Coll & A.K. Marks (Eds.), The immigrant
paradox in children and adolescents: Is becoming American a developmental
risk?, (pp. 253–274). Washingtion, DC: American Psychological Association.
Greenman, E., & Hall, M. (2013). Legal status and educational transitions of Mexi-
can and Central American immigrant youth. Social Forces, 91(4), 1475–1498.
Hagy, A. P. & Staniec, J. F. (2002). Immigrant status, race, and institutional choice in
higher education. Economics of Education Review, 21(4), 381–392.
534 The Review of Higher Education Summer 2015

Hall, M., Greenman, E., & Farkas, G. (2010). Legal status and wage disparities for
Mexican immigrants. Social Forces, 89(2), 491–513.
Heller, D. E. (1997). Student price response in higher education: An update to Leslie
and Brinkman. The Journal of Higher Education, 68(6), 624–659.
Kane, T., & Rouse, C. E. (1995). Labor-market returns to two- and four-year college.
American Economic Review, 85(3), 600–614.
Kaushal, N. (2008). In-state tuition for the undocumented: Education effects on
Mexican young adults. Journal of Policy Analysis and Management, 27(4),
771–792.
Light, A. (2001). In-school work experience and the returns to schooling. Journal of
Labor Economics, 19(1), 65–93.
Moretti, E. (2004). Estimating the social return to higher education: Evidence from
cross-sectional and longitudinal data. Journal of Econometrics, 121(1–2),
175–212.
National Conference of State Legislatures (2013). In-State Tuition and Unauthorized
Immigrant Students. Retrieved from http://www.ncsl.org.
National Immigration Law Center. (2009). Basic Facts about In-State Tuition for
Undocumented Immigrant Students. Retrieved from www.nilc.org.
National Immigration Law Center. (2013). State Bills on Access to Education for Im-
migrants. Retrieved from www.nilc.org.
Oropesa, R. S., & Landale, N. (2009). Why do immigrant youths who never enroll in
U.S. schools matter? School enrollment among Mexicans and Non-Hispanic
Whites. Sociology of Education, 82, 240–266.
O’Toole, D. M., Stratton, L. S., & Wetzel, J. N. (2003). A longitudinal analysis of the
frequency of part-time enrollment and the persistence of students who enroll
part time. Research in Higher Education, 44(5), 519–537
Passel, J. S. (2003). Further demographic information relating to the DREAM Act.
Washington, DC: Urban Institute.
Passel, J. S. (2005). Estimates of the size and characteristics of the undocumented
population. Washington, DC: The Pew Hispanic Center.
Passel, J. S. & Cohn, D. V. (2008). Trends in unauthorized immigration: Undocumented
inflow now trails legal inflow. Washington, DC: The Pew Hispanic Center.
Perez Huber, L., & Malagon, M. (2006). Silenced struggles: The experiences of La-
tina and Latino undocumented college students in California. Nevada Law
Journal: 841–61.
Perez, P. A., & Rodriguez, J. L. (2011). Access and opportunity for Latina/o undocu-
mented college students: Familial and institutional support Factors. Association
of Mexican American Educators Journal 5(1), 14–21.
Perna, L. W. (2006). Studying college choice: A proposed conceptual model. In J.C.
Smart (Ed.), Higher education: Handbook of theory and research. (pp. 99–157).
Netherlands: Springer.
Portes, A., & Rumbaut, R. G. (2006). Immigrant America: A portrait, 3rd edition.
Berkeley: University of California Press.
Potochnick, S. (2014). How states can reduce the dropout rate for undocumented
immigrant youth: The effects of in-state resident tuition policies. Social Sci-
ence Research, 45,18–32.
Darolia & Potochnick / Educational Implications of IRT Policies 535

Reich, G. & Barth, J. (2010). Educating citizens or defying federal authority? A com-
parative study of in-state tuition for undocumented students. Policy Studies
Journal, 38(3), 419–445.
Reich, G., & Mendoza, A. A. (2008). Educating kids’ versus ‘coddling criminals’:
Framing the debate over in-state tuition for undocumented students in Kansas.
State Politics and Policy Quarterly, 8(2), 177–197.
Shear, M. D. (2014, November 20). Obama, Daring Congress, Acts to Overhaul Im-
migration. New York Times. Retrieved from http://www.nytimes.com.
Stinebrickner, R. & Stinebrickner, T. (2003). Working during school and academic
performance. Journal of Human Resources, 38(3), 473–491.
Stratton, L.S., O’Toole, D.M, & Wetzel, J.N. (2008). A multinomial logit model of col-
lege stopout and dropout behavior. Economics of Education Review, 27, 319–331.
Suàrez-Orozco, C., Yoshikawa, H., Teranishi, R.T, & Suàrez-Orozco, M.M. (2011).
Growing up in the shadows: The developmental implications of unauthorized
status. Harvard Education Review, 81(3), 438–472.
Teranishi, R., Suarez-Orozco, C., & Suarez-Orozco, M. (2011). Immigrants in com-
munity colleges. The Future of Children, 21(1), 153–169.
Texas Higher Education Coordinating Board. (2011). Overview eligibility for in-state
tuition and state financial aid programs. Austin, TX. Retrieved from http://
www.thecb.state.tx.us
U.S. Bureau of Labor Statistics (BLS). (2013). College enrollment and work activity
of 2012 high school graduates. Washington, DC: United States Department of
Labor. Retrieved from http://www.bls.gov/news.release/hsgec.nr0.htm.
Wolfe, B., & Haveman, R. (2003). Social and nonmarket benefits from education in
an advanced economy. In Y. Kodrzycki (Ed.), Education in the 21st century:
Meeting the challenges of a changing world (pp. 97–131). Boston, MA: Federal
Reserve Bank of Boston.
Wooldridge, Jeffrey M. (2010). Econometric analysis of cross section and panel data.
Cambridge: MIT Press.
Reproduced with permission of the copyright owner. Further reproduction prohibited without
permission.