Professional Documents
Culture Documents
Does Aid Matter? Measuring the Effect of Student Aid on College Attendance and Completion
Author(s): Susan M. Dynarski
Source: The American Economic Review, Vol. 93, No. 1 (Mar., 2003), pp. 279-288
Published by: American Economic Association
Stable URL: http://www.jstor.org/stable/3132174
Accessed: 01-01-2016 19:14 UTC
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/
info/about/policies/terms.jsp
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content
in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship.
For more information about JSTOR, please contact support@jstor.org.
American Economic Association is collaborating with JSTOR to digitize, preserve and extend access to The American Economic
Review.
http://www.jstor.org
This content downloaded from 155.247.166.234 on Fri, 01 Jan 2016 19:14:10 UTC
All use subject to JSTOR Terms and Conditions
Does Aid Matter?
Measuringthe Effect of Student Aid
on College Attendance and Completion
By SUSANM. DYNARSKI*
The United States spends billions of dollars cent of full-time college studentsaged 18 to 21
each year on financial aid for college students, were receiving Social Security student
but there is little evidence that these subsidies benefits.1
serve their goal of increasingcollege attendance In 1981, Congressvoted to eliminatethe pro-
and completion. Determining whether aid af- gram. Enrollment sank rapidly (see Figure 1):
fects schooling decisions is an empirical chal- by the 1984-1985 academic year, program
lenge. The traditionalapproachhas been to re- spending had droppedby $3 billion. Except for
gress a person's educationalattainmentagainst the introductionof the Pell Grantprogramin the
covariates and the aid for which he is eligible early 1970's, and the various G.I. bills, this is
and interpretthe coefficient on aid as its casual the largest and sharpestchange in grant aid for
effect. However, this is problematic,as aid eli- college students that has ever occurred in the
gibility is correlated with many observed and United States. The program's demise provides
unobserved characteristics that affect school- an opportunityto measure the incentive effects
ing decisions. In order to identify the effect of financialaid. Using difference-in-differences
of aid, we need a source of variation in aid methodology, and proxying for benefit eligibil-
that is plausibly exogenous to unobservable ity with the death of a parentduringan individ-
attributesthat influence college attendance. A ual's childhood,I find thatthe eliminationof the
shift in aid policy that affects some studentsbut Social Securitystudentbenefit programreduced
not others is one such source of exogenous college attendanceprobabilitiesby more than a
variation. third. These estimates suggest that an offer of
In this paper,I analyze the impact on college $1,000 in grant aid increases the probabilityof
attendance and completed schooling of the attending college by about 3.6 percentage
elimination of the Social Security StudentBen- points. Aid eligibility also appears to increase
efit Programin 1982. From 1965 to 1982, the completed schooling.
Social SecurityAdministrationpaid for millions
of studentsto go to college. Underthis program, I. EmpiricalFramework
the 18- to 22-year-old children of deceased,
disabled, or retired Social Security beneficia- We are interested in the effect of aid on a
ries received monthly payments while enrolled person's educational attainment.This relation-
full time in college. The average annual pay- ship can be expressed with the following re-
ment in 1980 to the child of a deceased par- duced-formequation:
ent was $6,700. All dollar amounts are in real
terms ($2,000). At the program'speak, 12 per- (1) Si = + 3AIDi + XXi + si
This content downloaded from 155.247.166.234 on Fri, 01 Jan 2016 19:14:10 UTC
All use subject to JSTOR Terms and Conditions
280 THEAMERICANECONOMICREVIEW MARCH2003
3
Twenty-one percent of student beneficiaries were in
individual covariates. In order to obtain an un- high school (Phillip Springer,1976). Since I am focusing on
biased estimate of the causal effect of aid, the college-going behavior,I have excluded this groupfrom the
analyst must include in the regressionall of the figures in the text. Further,I do not find an effect of student
elements of Xi that affect both schooling and benefit eligibility on the high-school graduationrate(results
available from author).
aid eligibility. Unfortunately,many of these co- 4 Much of this
growth was due to rising disability and
variates are typically unavailableto the econo- retirement rates among prime-age men. Children of the
metrician.A discreteshift in the rules governing disabled and retiredwere 25 percentof studentbeneficiaries
aid eligibility can induce variation that is un- in 1965 and 40 percent in 1980.
5 Each child was
correlatedwith these unobserved determinants eligible to receive 75 percent of the
PrimaryInsuranceAmount (PIA), a nonlinear function of
of schooling. the deceased parent'searninghistory. If the individualben-
Traditionally,the Social Security Adminis- efits of a family summed to more than 175 percent of the
trationhas provided benefits to the children of PIA, each benefit was proportionallyreduced.
6
deceased, disabled, and retired Social Security Grant, loan, and tuition figures are from College
Board (1983). The schooling costs of beneficiaries were
beneficiaries only until those children are 18.2 furtherreduced by traditional student aid, which was only
But between 1965 and 1982, payments were minimally offset by Social Security student benefits. In
the federal aid formulas, a dollar in student benefits
reduced other aid by five cents. Colleges may have
treated student benefits less generously than this in cal-
2
For background on the program see Committee on culating their own scholarships. Such "crowd-out" of
Ways and Means (1979, 1982), Office of the Comptroller institutional aid would bias the paper's estimates toward
General (1979), and Rebecca Luzadis (1983). zero.
This content downloaded from 155.247.166.234 on Fri, 01 Jan 2016 19:14:10 UTC
All use subject to JSTOR Terms and Conditions
VOL.93 NO. I DYNARSKI:DOES AID MATTER? 281
In 1981, Congress voted to eliminate the So- cation controlsfor changes over time in average
cial Security studentbenefit program.Those not college attendanceratesand averagedifferences
enrolled in college as of May 1982 were ineli- in the college attendance of those with a de-
gible for future subsidies, while those currently ceased fatherand those with a living father.The
enrolled in college had their payments severely key identifying assumption is that any relative
reduced.By the 1983-1984 academicyear, pro- shift in the attendance of the children of de-
gram spending had dropped to $0.38 billion. ceased fathers is attributableto the withdrawal
This sharp policy shift creates variation in aid of student benefits. Note that /3 captures the
eligibility that can be used to identify the effect effect on schooling decisions of aid eligibility
of grant aid on schooling decisions. rather than aid receipt. In the language of the
I use difference-in-differencesmethodology experimentalistliterature,/3capturesthe effect
to examine the effect of eligibility for Social of the intention to treat. Since policy makers
Security student benefits on college attendance control the offer of aid, but not its take-up, /3 is
and completed schooling. The key estimating the parameterof interest if we wish to predict
equation is the following: the effect of altering aid policy.
The dataused in the analysis are the National
(2) LongitudinalSurvey of Youth (NLSY). See the
Data Appendix for details. Survey respondents
Yi = a + f3(Father Deceasedi X Beforei) in their senior year of high school in the springs
of 1982 and 1983 form the "after"cohorts, who
+ 6Father Deceasedi were ineligible for any student benefits upon
high-school graduation. Those who were se-
+ OBeforei + Vi niors in the springsof 1979, 1980, or 1981 form
the "before"cohorts.
where the dependent variable is a measure of Means, shown in Table 1, are presentedsep-
college attendanceor completionand Beforei is aratelyfor the periods before and after the pol-
a binary variable that is set to one if a youth is icy change and for those with living and
a member of a cohort that graduatedfrom high deceased fathers.Five percentof childrenhad a
school before studentbenefits were eliminated. male parentdie before the child turned 18; this
Father Deceasedi is a binary variable set to figure is consistent with mortalitytables for the
one for those who, due to the death of their relevant age cohorts. Children with deceased
father,were potentiallyeligible for child survivor fathers grow up in relatively low-income fami-
benefits.7I focus on eligibility due to the death lies and are more likely to live in single-parent
of the parent because parental disability and households.9Childrenwith deceased fathersare
retirementmay be endogenously determinedby more likely to be black, due to the higher mor-
the availability of studentbenefits. This endog- tality rate of prime-age black men, and they
enous selection into aid eligibility would bias have lower Armed Forces Qualification Test
upward the difference-in-differencesestimator. (AFQT) scores.10 As the last column shows,
I focus on fathersbecause 90 percentof student these differences between the two groups are
beneficiaries were entitled to benefits through stable over time. During the period under anal-
their fathers.8 ysis, there is no statistically significant change
The reduced-form effect of Social Security in the differences in backgroundcharacteristics
student benefits is capturedby /3. The specifi- between children with living and dead fathers.
9 Some wives
7Having a deceased father is an imperfect proxy for remarryafter the husband'sdeath, and so
benefit eligibility, since some fathers did not work long not all deceased-father children live in a single-parent
enough in covered employment to generate survivor bene- household. Children continue to be eligible for survivor
fits. As I discuss in Section III, this mismeasurementof benefits if their mothers remarry.
eligibility is minor and will bias the paper's estimates to- 10Blacks are twice as
likely to have a dead father (7.2
ward zero. percentvs. 3.4 percent).This is consistent with the National
8See Social Security Administration(1982). Fathersare LongitudinalMortality Survey, which shows that the mor-
more likely to have a sufficient working history to generate tality rate for black men aged 25 to 50 is twice that of white
survivor benefits upon death. men.
This content downloaded from 155.247.166.234 on Fri, 01 Jan 2016 19:14:10 UTC
All use subject to JSTOR Terms and Conditions
282 THEAMERICANECONOMICREVIEW MARCH2003
TABLE1-NLSY SUMMARY
STATISTICS
Notes: Means are of NLSY poverty and random samples, weighted by 1988 sample weights. Income and household
composition measured during senior year of high school. AFQT is age adjusted;see Data Appendix. Standarderrors are
in parentheses. Standard errors in the difference-in-differences column adjusted for clustering at the household
level.
This content downloaded from 155.247.166.234 on Fri, 01 Jan 2016 19:14:10 UTC
All use subject to JSTOR Terms and Conditions
VOL.93 NO. I DYNARSKI:DOES AID MATTER? 283
This content downloaded from 155.247.166.234 on Fri, 01 Jan 2016 19:14:10 UTC
All use subject to JSTOR Terms and Conditions
284 THEAMERICANECONOMICREVIEW MARCH2003
seen in Table 1, the childrenof deceased fathers TABLE 3-OLS, EFFECT OF ELIGIBILITY FOR
STUDENT BENEFITS ON SCHOOLING OUTCOMES
were more likely that their counterpartsto at-
BY AGE 23 AND AGE 28
tend college before the policy change but less
likely to attend college afterward.Under these Attended Completed
conditions, linear and nonlinear analysis will college any years Years of
produceestimatesof the same sign, thoughtheir full time of college schooling
magnitudemay vary. By age 23 0.219 0.145 0.564
The results in Table 2 suggest that aid eligi- (0.102) (0.123) (0.379)
bility has a strong effect on college attendance. By age 28
In the next section, I will put the magnitudeof Lower bound 0.224 0.178 0.679
(0.106) (0.113) (0.399)
this effect in context. I firstexamine whetheraid Exclude attriters 0.248 0.191 0.754
eligibility increasedcompleted schooling in ad- (0.111) (0.118) (0.408)
dition to college entry. These estimates are of Assign last value 0.256 0.211 0.727
interestbecause it is completedschooling thatis (0.105) (0.112) (0.397)
rewardedby the labor market, rather than at- EstimatesAdjustedfor ClassificationError
tempted schooling. If the marginal college en-
By age 23 0.243 0.161 0.626
trantis not capableof completingeven a year of By age 28
college, then the attendance results discussed Lower bound 0.249 0.198 0.754
above will substantially overstate the social
benefits of studentaid. The estimates, based on Notes: Coefficientsare those on deceased father X before in
the fully controlledspecificationof Table 2, are regressionsin which the outcomes are those indicatedin the
columns. The regression specificationis that of column (2)
in Table 3. Eligibilityfor studentbenefitsappears in Table 2. See text for explanation.
to increasethe probabilityof completingat least
a year of college by 14.5 percentagepoints and
years of completed schooling by about half a For none of the outcomes are the estimates at
year, though neither estimate is significant. age 28 lower thanthose at age 23. This suggests
The positive effect of aid eligibility on atten- that aid eligibility did not simply speed up in-
danceandcompletioncould dissipateover time if vestmentin schooling but also raisedits optimal
studentbenefitsinduce studentsto simply accel- level. The three approachesto handlingattrition
erate,ratherthanincrease,theirschoolinginvest- yield similar results. The lower-boundestimate
ments.I thereforeexamineschoolingdecisionsas is that eligibility for student benefits increases
of age 28. Thereis attritionbetween age 23 and the probabilityof attendingcollege by age 28 by
28; nonrandomattritionhas the potentialto bias 22.4 percentagepoints, which is almost identi-
these estimates.15 I test three alternative ap- cal to the effect estimated at age 23. If we
proaches to dealing with attrition,all of which examine the other schooling outcomes, the
yield similarresults:I dropthe attriters,I assign same patternemerges: the lower bound of the
them their last observed value of the dependent effect at age 28 is just slightly above the effect
variable,and I assign them values thatprovide a at age 23, indicatingthat the effect of aid eligi-
lower boundon the effect of aid eligibility in the bility does not dissipate over time. If anything,
presence of nonrandomattrition.I estimate this it appearsthatthe effect on completedschooling
lower bound by imputing to attritersschooling rises over time, though the size of the standard
values that would be produced by a negative errorsprecludes any strong conclusions.
correlation between the aid eligibility and
schooling.16Results are in Table 3. III. Discussion and Conclusions
This content downloaded from 155.247.166.234 on Fri, 01 Jan 2016 19:14:10 UTC
All use subject to JSTOR Terms and Conditions
VOL.93 NO. I DYNARSKI:DOES AID MATTER? 285
otherwise have entered. While many of these duces an increase of 3.6 percentage points in
marginalentrantswill completejust a few years the share of high-school graduates attending
of college, aid likely induces others who would college.20
have completed just a few years of college to The three-yearphaseout of benefits that be-
instead finish their degrees. The effect of move- gan in 1982 complicatesthe interpretationof the
ment along these margins is a relative increase completion results. A beneficiary graduating
by age 28 in average schooling of 0.679 years. from high school in 1981 would have expected
Classification error will bias these estimates four years of full student benefits, but would
toward zero. Misclassification of the eligible have instead received a subsidy that declined
group is minor:the Social SecurityAdministra- from $6,700 in the freshmanyear to $1,100 in
tion estimates that, in the early 1980's, 95 per- the senior year. On average, members of the
cent of children under 18 would have been before cohorts who stayed in college for four
eligible for survivor benefits had a working years would have received annual benefits of
parent died. Misclassification of the control $4,700. Therefore,each $1,000 of studentben-
group is also minimal. The share of 17-year- efits offered induced an increase of 0.16 years
olds in 1980 whose fathers were not dead but (=0.754/4.7) in the completed schooling of
were eligible for Social Security due to the high-school graduatesattendingcollege.2'
disability or retirement of a parent was 5.3 How do these results compare with previous
percent.17This degree of misclassificationindi- estimates of the effect of aid on schooling de-
cates that the estimates of Table 3 should be cisions? LarryLeslie and Paul Brinkman(1988)
adjusted upward by 11 percent.18The bottom review several dozen college attendancestud-
panel of Table 3 contains these adjusted esti- ies, which generally suggest that a $1,000 de-
mates, which suggest that aid eligibility in- crease in net price is associated with a 3- to
creases the probabilitiesof attendingcollege by 5-percentage-pointincrease in attendance.With
age 23 by 24.3 percentage points and of com- few exceptions, discussed below, these esti-
pleting at least a year of college by 16.1 per- mates are vulnerable to the biases discussed
centage points. The effect on completed earlierin the paper. Several studies that control
schooling at age 28 is 0.754 years. for unobserveddeterminantsof schooling have
These are large effects, but so too was the examined the effect of aid on college entry. By
financialincentive. The average studentbenefit contrast, none have examined its effect on the
for the child of a deceased parent was about completed schooling of young people, and so
$6,700 in 1980, more than enough to cover the the presentcompletion results, while imprecise,
$1,900 cost of tuition and fees at a public uni- break new ground.22
versity. If we sum the direct and opportunity Using within-statevariationin public tuition
costs of college, the latterproxied by the annual costs, Thomas J. Kane (1994) concludes that a
wage of young high-school graduates($18,500 $1,000 dropin tuitionproducesa 3.7-percentage-
in 1980) we obtain an elasticity of attendance point increase in college attendance.Dynarski
with respect to schooling costs of about 1.5.19
Each $1,000 of student benefits offered in-
20
The effect of aid may be nonlinear.In the presence of
liquidity constraints, a threshold amount of aid may be
17
The numberof 17-year-oldsis from the 1980 Census needed to affect behavior, leading a large grant to have a
and the number of 17-year-olds with parents retired or on largerper-dollareffect thana small grant.It is also plausible
disability is from Table 54 in Social Security Administra- that the marginaleffect of aid falls as aid rises.
tion (1982). Note that the NLSY contains no information 21 The interpretationof the attendance results is not
about parents' retirementor disability status, so these indi- complicatedby the phaseout.The college entry decision of
viduals cannot be identified. See Dynarski (1999) for an those affected by the phaseout occurredunderthe assump-
analysis thatuses father's age to proxy for benefiteligibility tion of full benefits, since 85 percentwho enteredcollege by
due to father's retirementstatus. age 23 did so directly after high school.
18 See Dennis Aigner (1973) and Richard Freeman 22
The only comparablecompletion studies have focused
(1984) for the derivationof this correctionfor classification on the effect of the G.I. bills on the schooling of veterans,
error. whose behavior is likely quite different from that of the
19Annual wage is the average weekly wage of 19-year-
typical young person. JoshuaAngrist (1993), Marcus Stan-
old high-school graduatesin the Merged OutgoingRotation ley (2000), and John Bound and Sarah Turner (2002) all
Group of the CurrentPopulation Survey multiplied by 50 conclude that veterans' education benefits have a positive
and inflated to currentdollars. effect on completed schooling.
This content downloaded from 155.247.166.234 on Fri, 01 Jan 2016 19:14:10 UTC
All use subject to JSTOR Terms and Conditions
286 THEAMERICANECONOMICREVIEW MARCH2003
(2000) finds that Georgia's HOPE Scholarship, the deceased parentand its transactioncosts are
a merit-aid program, produced an increase in extremely low.26 In traditional programs, the
attendanceof 4 percentagepoints per $1,000 of correlation between aid and parents' human
aid offered. W. Lee Hansen (1983) and Kane capital is negative. The paper's estimates will
(1995) find that the introduction of the Pell provide a biased prediction of the effect of
Grant had no effect on college attendance.23 need-based aid if the response to aid is hetero-
Neil Seftor and Turner(2002) find an effect of geneous and correlated with parental human
Pell Granteligibility on the attendancerate of capital. If children from high-human-capital
older adults;they estimate an effect of 0.7 per- families are most sensitive to cost, the Social
centage points per $1,000 in aid.24In summary, Security programwill channel dollars to high-
with the exception of the Pell studies, estimates response individualsand the paperwill overes-
that do and do not account for unobservable timatethe effect of traditionalaid. The opposite
differences across individuals reach similar will hold if, instead,childrenfrom high-human-
conclusions: a $1,000 drop in schooling costs capital families are less sensitive to cost.27 The
increases college attendanceby 3 to 4 percent- program'svery low transactioncosts unambig-
age points. This suggests that either the cross- uously tend to make the present estimates pro-
sectional results are unbiased,or, as is the case vide an upper bound on the incentive effect of
in the return-to-schoolingliterature,competing traditionalstudent aid, which imposes substan-
biases cancel in a cross-sectional analysis. tial transactioncosts on applicants.
Are the present estimates informative as to From a student's perspective, this program
the effect of traditional aid, such as the Pell made college the optimal choice at very low
Grant?Are the studentbenefit programand the rates of return to schooling. With the student
populationit served sufficientlyunique that the benefit, a year of college would pay for itself
estimates' external validity is extremely con- with a rate of returnas low as 2.5 percent.28The
strained?The youth that are the focus of this program therefore likely induced into college
paper are special in at least one way: their some studentswith a very low ex ante payoff to
fathers are dead. This may make their families schooling. But given the rapidrise in the return
especially sensitive to the price of college, since to schooling since the early 1980's, ex post
they have only one parent's labor supply with returns for those induced into college by the
which to buffer shocks to schooling costs. On studentbenefit were likely considerablyhigher.
other observable characteristics,the population Do the paper'shigh estimatesof the elasticity
eligible for student benefits closely resembles of college attendancewith respect to aid indi-
that served by need-basedprograms.In partic- cate the presence of liquidity constraints?Since
ular, both groups are disproportionatelyblack grant aid reduces the cost of schooling and
and from low-income families.25 therebyincreases its optimal level, a behavioral
In its structure,the Social Securityprogramis
unusualin thatbenefits rise with the earningsof
26
Child beneficiaries had to do little to obtain student
benefits. The Social Security Administrationsent form let-
23
Sarah Turner(2000) suggests that schools may have ters to child beneficiariesnearingage 18. To those respond-
"undone"the Pell Grant by lowering the institutional aid ing that they would be continuing their schooling, SSA
they offered low-income students, thereby explaining the mailed separate, monthly benefit checks until the benefi-
zero programeffects. ciary left school, marriedor turned 22. Schools provided
24 While this effect is smaller than the present estimate, annualverificationof enrollmentto the government.
27
it is comparablein magnitudeonce the estimates are scaled Dynarski(2000) shows that,in a simple human-capital
by the baseline share of the relevant population that is in framework,the sign of the correlationis ambiguous. Kane
college. (1994) presentsresults that indicate that low-income youth
25
Twenty-six percentof deceased-fatherchildrenand 21 are most sensitive to tuition.By contrast,Stanley (2000) and
percent of Pell Grantrecipients are black (National Center Turnerand Bound (forthcoming)find that the post-World
for EducationStatistics, 2000). The average family income War II G.I. Bill had the greatest impact on veterans from
in 1973 of studentbeneficiariesplaced them in the bottom more privileged backgrounds.
28 This calculation assumes costs consisting of tuition
income quartileof families with children in college, while
90 percent of dependent Pell Grant recipients are from and fees at a public university($1,900 in 1980-1981) and a
families with incomes below $40,000 (Bureauof the Cen- year of forgone earnings($18,500). I furtherassume a work
sus, 1974; Springer, 1976; National Center for Education life throughage 65, annualreal wage growth of 1 percent,
Statistics, 1998). and a real discount rate of 4 percent.
This content downloaded from 155.247.166.234 on Fri, 01 Jan 2016 19:14:10 UTC
All use subject to JSTOR Terms and Conditions
VOL.93 NO. 1 DYNARSKI:DOES AID MATTER? 287
This content downloaded from 155.247.166.234 on Fri, 01 Jan 2016 19:14:10 UTC
All use subject to JSTOR Terms and Conditions
288 THEAMERICANECONOMICREVIEW MARCH2003
This content downloaded from 155.247.166.234 on Fri, 01 Jan 2016 19:14:10 UTC
All use subject to JSTOR Terms and Conditions