Merit Aid, College Quality and College Completion:
Massachusetts’ Adams Scholarship as an In-Kind Subsidy∗
Sarah Cohodes
Harvard University
cohodes@fas.harvard.edu
Joshua Goodman†
Harvard University
Joshua Goodman@hks.harvard.edu
January 31, 2014
Abstract
We analyze a Massachusetts merit aid program that gives high-scoring students tuition
waivers at in-state public colleges with lower graduation rates than available alternative colleges. A regression discontinuity design comparing students just above and below the eligibility threshold finds that students are remarkably willing to forgo college quality and that
scholarship use actually lowered college completion rates. These results suggest that college
quality affects college completion rates. The theoretical prediction that in-kind subsidies of
public institutions can reduce consumption of the subsidized good is shown to be empirically
important.
∗
For making the data available for this project, we are indebted to Carrie Conaway, Director of Planning, Research
and Evaluation, and Robert Lee, MCAS Chief Analyst, both at the Massachusetts Department of Elementary and Secondary Education, as well as Jon Fullerton, Executive Director of the Center for Education Policy Research at Harvard
University. We also thank for helpful comments Chris Avery, David Deming, Sue Dynarski, David Figlio, Ed Glaeser,
Sandy Jencks, Tom Kane, Kevin Lang, Jeff Liebman, Bridget Long, Dick Murnane and Marty West, as well as seminar
participants at Harvard, MIT, Boston University, NBER, the Federal Reserve Banks of Boston and New York, APPAM,
AEFP and SOLE. Institutional support from the Harvard Kennedy School of Government and Taubman Center for State
and Local Government are gratefully acknowledged. Any errors are our own.
†Corresponding author.
1 Introduction
Recent research has emphasized troubling trends in U.S. college completion rates over the past few
decades. Among students entering college, completion rates are lower today than they were in the
1970s, due largely to low completion rates of men and students from lower socioeconomic backgrounds (Belley and Lochner 2007, Bailey and Dynarski 2011). This trend has spurred a vigorous
debate over the relative importance of factors that vary across students, such as academic skill and
family financial resources, and factors that vary across postsecondary institutions, such as funding
levels or management quality. Distinguishing the influence of student-level and institution-level
factors on college completion rates is confounded by the non-random selection of students into
institutions of different apparent quality. In this paper, we provide further evidence consistent
with the fact that the quality of the institutions themselves affects college completion rates.
To do so, we exploit is a Massachusetts merit aid program in which high school students with
test scores above multiple thresholds were granted tuition waivers at in-state public colleges. Such
colleges were of lower quality than the average alternative available to such students, where quality is measured by a combination of graduation rates, academic skill of the student body, and
instructional expenditures, as suggested by Black and Smith (2006).
The scholarship, though relatively small in monetary value, induced substantial changes in college choice, allowing us to
estimate the impact of college quality on students’ postsecondary enrollment decisions and rates
of degree completion. A regression discontinuity design comparing students just above and below the eligibility threshold finds that students are remarkably willing to forgo college quality for
relatively little money and that marginal students lowered their college completion rates by using
the scholarship. College completion rates decreased only for those subsets of students forgoing
the opportunity to attend higher quality colleges when accepting the scholarship. We describe
the magnitude of this response as remarkable because the value of the scholarship is dwarfed by
estimates of the forgone earnings of attending a lower quality college or failing to graduate. As
a whole, these results suggest that college quality has a substantial impact on college completion rates. We also find clear evidence that this scholarship increased college enrollment, though
not graduation, rates for the most disadvantaged students.
Such students comprise, however, an
1
extremely small fraction of the total pool of those eligible for this merit aid.
Our research contributes to three strands in the literature on postsecondary education and the
public subsidy of such education. First, a now extensive literature documents the sensitivity of
students’ college enrollment decisions to financial aid generally (Deming and Dynarski 2010, Kane
2006) and merit aid more specifically (Dynarski 2000, Cornwell et al. 2009, Dynarski 2008, Kane
2007, Pallais 2009, Goodman 20081
). In contrast to most of the programs studied in this literature,
the Adams Scholarship targets a very highly skilled set of students, namely the top 25% of high
school graduates in each school district. As a result, our estimates are generated by a part of the
skill distribution not often studied. Furthermore, unlike in most aid programs, recipients were
automatically notified of their eligibility without having to apply. Simplifying the aid process is
known to affect students’ college enrollment decisions (Bettinger et al. 2012),
so that this program
design may explain in part the large impacts of aid observed here. Our results are also consistent
with Fitzpatrick and Jones (2012), which finds that merit aid does effectively keep some students
in state but that marginal students are a small fraction of total aid recipients.
Second, we add to the growing literature on the impact of college quality on student outcomes.
Much of the literature on the impact of college quality on degree completion has focused on the
community college sector, reaching varying conclusions about whether access to and quality of
community colleges affects educational attainment (Rouse 1995, Leigh and Gill 2003, Sandy et al.
2006, Calcagno et al. 2008, Stange 2009, Reynolds 2012). Estimates of the impact of college quality on labor market earnings are similarly varied, with some positive (Loury and Garman 1995,
Brewer et al. 1999, Chevalier and Conlon 2003, Black and Smith 2004, Black and Smith 2006, Long
2008, Hoekstra 2009, Andrews et al. 2012), some zero or positive only for disadvantaged subgroups (Dale and Krueger 2002, Dale and Krueger 2011), and some suggesting that earnings differences dissipate once the job market properly understands graduates’ underlying ability (Brand
1This paper represents an extension and improvement of Goodman (2008), which studied the same merit aid program at an earlier time and used less informative outcome data. In particular,
that earlier paper could only measure
whether graduating high school seniors self-reported their intention to enroll in public or private colleges, without
identification of specific campuses, actual enrollment or persistence and graduation. This paper, in contrast, uses substantially more detailed administrative data that allows identification of the specific institutions students actually enroll
in, as well measurement of persistence and graduation rates. This allows for clear estimation of the quality and cost
tradeoffs students are making, the impact of this merit aid on in-state enrollment and, perhaps most importantly, the
impact of this aid on college graduation rates.
2
and Halaby 2006, Lang and Siniver 2011). Nearly all of these research designs attempt to eliminate
selection bias either by conditioning on students’ observable characteristics or by instrumenting
college quality with distance from or tuition of nearby colleges. Neither approach entirely eliminates the possibility that unobserved student-level factors may be driving their estimates.
The
exception to this is Hoekstra (2009), which uses a discontinuity inherent in the admissions process
to a flagship university to estimate the labor market return to an elite college education. We employ a similarly identification strategy and unlike Hoekstra are able to observe the college choice
made by students not enrolling in the target institutions, allowing us to estimate the impact of
merit aid on college quality. Though sources of exogenous variation in school and curriculum
quality are more common at lower levels of schooling because of school choice lotteries (Deming et al. 2011) and test score-based admissions rules (Bui et al. 2011, Abdulkadiroglu et al. 2011,
Dobbie and Fryer 2011), they are rarer in the postsecondary literature.
Furthermore, our finding that college quality plays an important role in completion rates is
consistent with important pieces of recent evidence. Controlling for rich sets of student characteristics does not eliminate wide variation among postsecondary institutions in completion rates
(Bowen et al. 2009).
Students who attend college in large cohorts within states have relatively
low college completion rates, likely stemming from decreased resources per student given states’
tendencies to change public postsecondary budgets slowly (Bound and Turner 2007). Bound et al.
(2010) argue that the vast majority of the decline in completion rates can be statistically explained
by decreasing resources per student within institutions over time and, even more importantly,
shifts in enrollment toward the relatively poorly funded public sector. All of this suggests that
characteristics of colleges themselves, such as resources available per student, play an important
role in completion rates and that student-level factors are only part of the story.
Third, we show the empirical importance of the theoretical possibility first discussed in Peltzman (1973) that in-kind subsidies of public institutions can reduce consumption of the subsidized
good. Prior work has shown how public in-kind subsidies can generate at least partial crowdout of privately provided health insurance (Cutler and Gruber 1996, Brown and Finkelstein 2008),
preschools (Bassok et al. 2012) and two-year colleges (Cellini 2009). Peltzman’s contribution was
3
the prediction that, in some cases, crowdout could theoretically be large enough to reduce overall consumption of the subsidized good.
Work by Ganderton (1992), using cross-state variation
in tuition subsidies, and Long (2004), using much finer college-specific variation in such subsidies, suggests that this in-kind public support for postsecondary education does reduce overall
spending on education. We contribute to this literature by providing the first evidence of such
reduced consumption driven by an exogenous shock in the size of the in-kind subsidy. We also
show that this reduced spending on higher education comes at the cost of a reduced probability
of degree completion, a possibility recognized by Kane (2007) in his evaluation of the D.C. Tuition Assistance Grant program but unexplored because too little time had passed to look beyond
enrollment effects.
The structure of the paper is as follows. In section 2, we describe the merit scholarship program in detail. In section 3, we describe the data on students and colleges, including our measures
of college quality.
In section 4, we explain our empirical strategy, a regression discontinuity design that accounts for the multiple thresholds students must cross in order to be eligible for aid.
In section 5, we present estimates of the impact of college quality on enrollment decisions and
completion rates. In section 6, we discuss implications of our findings and conclude in section 7.