In summary, the primary effect of the Adams Scholarship was to induce large numbers of
students to switch into in-state public four-year colleges from other four-year colleges they otherwise would have attended, a result consistent with Goodman (2008). The scholarship did increase
in-state college enrollment rates but had little impact on in-state graduation rates. Scholarship
eligibility actually reduced overall graduation rates, for reasons we now turn to.
5.2 College Quality and Cost
The most plausible explanation for the negative impacts on graduation rates is that the scholarship
induced students to attend colleges with substantially lower graduation rates than they otherwise
would have. Table 6 explores the quality and cost tradeoffs that the Adams Scholarship induced.
The top row presents reduced form estimates of the impact of scholarship eligibility on a variety
of college quality and cost measures, as in Equation 1 above. For this analysis,
we assign students
to the four-year college to which they enroll immediately following high school graduation. The
bottom row estimates these impacts for the marginal student using the following equations that
instrument enrollment in an in-state public college with scholarship eligibility:
Yijt = β0 + β1AdamsCollege ˆ
ijt + β2Gapijt + β3Gapijt × Adamsijt + �ijt (2)
AdamsCollegeijt = α0 + α1Adamsijt + α2Gapijt + α3Gapijt × Adamsijt + νijt (3)
In the first column, we generate an indicator for a college being highly competitive if Barron’s 2009
rankings placed that college into one of its top three categories of “most competitive,” “highly
competitive,” and “very competitive,”. None of Massachusetts’ public colleges fall into these categories, which include colleges such as Boston University, Tufts University, Simmons College, and
Lesley University. All of the U. Mass. campuses and nearly all of the state colleges fall into the
fourth category of “competitive,” which also includes private colleges such as Suffolk University
and the Wentworth Institute of Technology. The fifth category of “not competitive” includes two
state colleges and all community colleges. Column 1 shows that, for the classes of 2005-06, scholarship eligibility induced an estimated 3.3% of students, or 48% of those switching colleges, to forgo
20
institutions in those highest three categories. Students did not simply switch into the public sector from private or out-of-state colleges of similar quality. Half of the students induced to switch
colleges would have enrolled in more competitive alternatives in the absence of the scholarship.
Other measures of college quality, which are defined only for students immediately enrolling
in four-year colleges, point to a similar pattern. In column 2, the estimates suggest that students
induced by the scholarship to switch into Adams colleges would otherwise have attended colleges with four-year graduation rates nearly 17 percentage points higher. These marginal students
would also have attended colleges with higher SAT math scores (by 27 points) and higher instructional spending per student (by $3,700 annually), though this last estimate is not statistically
significant. Combining these three measures as described above, column 5 shows that scholarship eligibility induced the marginal student to forgo more than 0.6 standard deviations in college
quality.
In exchange for this drop in quality, students enrolled in public colleges where the average
student’s net price of attendance was $10,000 a year lower than private and out-of-state alternatives, as seen in column 6. This is the most direct measure we provide of the extent to which this
in-kind subsidy reduces consumption of college education, as in Peltzman (1973). This cost difference would, however, have been available to these students even in the absence of the Adams
Scholarship. The scholarship itself was worth, on average, less than $1,400 a year to such students,
as seen in column 7.27 Combining the estimates from columns 5 and 7 suggests a willingness to
forgo 0.47 (0.637/1.353) standard deviations of college quality per $1,000 in annual aid. Estimates
for the larger sample of 2005-08 graduates are quite similar. Below, we calculate the estimated
impacts of such a tradeoff on lifetime earnings and find that these responses are hard to explain
in classical human capital model. Students seem remarkably willing to forgo college quality and
attend institutions with low graduation rates in exchange for relatively small amounts of financial
aid.
To strengthen our case that the decrease in college quality induced by the scholarship explains
the observed graduation impacts, we explore heterogeneity by a variety of characteristics in Table
27This is a weighted average of enrollment across all of the in-state public four-year colleges, where the value of the
scholarship ranged from $1,417-$1,714 at U. Mass. campuses and $910-$1,030 at state colleges.
21
7. Here we use the classes of 2005-08 to improve precision among relatively small subgroups.
Panel A takes advantage of the fact that the academic skill level defined by eligibility threshold
varied by school district due to the requirement that students be in the top 25% of their district
peers. We therefore divide districts into quintiles by the fraction of 2004 graduates who attended
a competitive college, as defined previously. We then fully interact our baseline specification from
prior tables with indicators for being from the bottom quintile, middle three quintiles, and top
quintile of school districts, and also include the direct effects of those indicators.
For students from the bottom quintile districts, who are on average lower income and less academically skilled, scholarship eligibility increases enrollment in Adams colleges by nearly eight
percentage points, of which five percentage points represent students who would not otherwise
have enrolled immediately in any four-year college. Eligibility does not, however, reduce such students’ probability of attending a highly competitive college likely because they would not have
applied to or gained admission to such institutions. In the absence of the scholarship, these students would be attending four-year colleges of similar quality to the Adams colleges, or none at
all. For these students, eligibility has no statistically significant impact on persistence and graduation on persistence, with point estimates that are slightly positive. The Adams Scholarship
substantially increases college enrollment for such students, has little impact on college quality,
and little impact on persistence and graduation.
Students from the top quintile districts barely react to this aid at all. There is no statistically
significant evidence that they enroll in Adams colleges at higher rates, forgo highly competitive
colleges, or persist or graduate at lower rates as a result of scholarship eligibility. Such students do
not react presumably because they are wealthier on average and because their alternative college
options are so much higher quality than the Adams colleges that the scholarship is insufficient
incentive to switch.
Students from the middle quintile districts do react strongly to the aid. Eligibility raises enrollment in Adams colleges by nearly seven percentage points.
Little or none of this comes from
students enrolling in four-year colleges who would not have otherwise. Strikingly, two-fifths
(.026/.068) of those marginal students who switch into Adams colleges do so by forgoing highly
22
competitive colleges, unlike students from bottom or top districts. And, unlike students from bottom or top districts, only these students have clearly lower persistence and graduation rates as a
result of scholarship eligibility. Unlike students from the lowest scoring districts, who are induced
by the Adams Scholarship to switch sectors but not college quality level and have no graduation
effects, students from middle districts are induced to forgo college quality and are the only students whose graduation rates suffer. This strengthens the case that college quality explains the
scholarship’s negative graduation effect.
The remaining three panels in Table 7 explore heterogeneity by student race/ethnicity, poverty
and gender. Scholarship eligibility has a substantially stronger impact on non-white students’
enrollment decisions than on white students’ enrollment decisions. For non-white students, eligibility increases enrollment in an Adams college by over 13 percentage points, half of whom
are students who would not otherwise have enrolled in a four-year college at all. For white students, eligibility increases Adams college enrollment by only five percentage points, nearly all of
whom would have otherwise have enrolled in another four-year college. White students are the
only ones who forgo more competitive colleges and see their persistence rates suffer. Non-white
students do not forgo more competitive colleges and do not see a drop in their persistence rates.
Roughly similar patterns are seen with poor and non-poor students.
There are no statistically significant differences by gender. Taken as a whole, these results suggest that the observed negative
impacts on graduation rates are concentrated in subgroups of students who are induced to forgo
college quality by scholarship eligibility.
Having shown that scholarship eligibility both induced students to forgo college quality and
lowered their graduation rates, in Table 8 we directly estimate the impact of college quality on
those graduation rates. For such estimates to be valid, the exclusion restriction must hold, namely
that scholarship eligibility affects graduation rates only through the college quality channel. We
consider two potential violations of this exclusion restriction. First, scholarship eligibility may
affect not only marginal students but infra-marginal ones as well. Our estimates suggest that
roughly three-fourths of scholarship users would have attended Adams colleges in the absence
23
of the scholarship.28 If the financial aid were changing their graduation rates, the IV estimates
would confound that channel with the quality channel. We believe this is unlikely both because
the amount of money involved here is small relative to the costs of college and because that small
amount of additional aid should, if anything, help students graduate by allowing them not to
work while on campus. If the graduation rates of infra-marginal students were improved by
this aid, the coefficients below would actually underestimate the impact of college quality on the
graduation rate of marginal students.
