AFAWG Financial Aid Budget Control

Recommendations
February 21, 2017 SUMMARY DRAFT FOR DISCUSSION

Admissions and Financial Aid Working Group: Barbara Knuth (Chair),

Yamini Bhandari (Student Assembly/Trustee), Vicki Bogan, Kathryn Boor,
Marin Clarkberg, Lance Collins, Ron Ehrenberg, Kevin Hallock, Sue Hitchcock,
Bill Lesser, Jason Locke, Ryan Lombardi, A.T. Miller, Alan Mathios, Mitchell
McBride/Akhilesh Issur (Student Assembly), Dan Robertson, Gretchen Ritter,
Laura Spitz, Paul Streeter, Rotating Student Member (Student Assembly)

Introduction

The main priorities for the Admissions and Financial Aid Working Group
(AFAWG) in 2016-2017 focused on:
(1) Rebalancing: considering the extent to which undergraduate
financial aid programs fairly address need across socioeconomic groups, with
a particular focus on whether there should be a rebalancing of how financial
aid resources are distributed; and
(2) Sustainability: identifying and evaluating possible options for cost-
savings for the undergraduate financial aid program while continuing
Cornells commitment to financial aid providing access for all qualified
students should there be another significant economic downturn that creates
a need to control spending on Cornell grant aid.

The recommendations below from the AFAWG are intended as a suite of

choices for the university leadership to consider and choose from, depending
on the enrollment priorities and financial aid budget capacity for the
university. We include a few options for modest increases in grant aid
expenditures to shift loan maximum cut-off points to correspond more
closely to current US income quintiles in the interest of addressing the
fairness question particularly for middle and upper US income quintiles,
and we describe several options that could be considered if the university
needs to control grant aid expenditures in anticipation of or response to a
significant economic downturn that would otherwise cause an increase in
grant aid costs.

The university leadership could decide to implement some of the cost-

savings options to liberate grant aid that could then be applied to meet the
additional grant aid expenditures associated with shifting upward the loan
maximum cut-off points, essentially holding overall grant aid expenditures
steady but redistributing the expenditures among income groups. However,
the AFAWG examined extensive socioeconomic data related to Cornell
applicants, admits, and enrollment and concluded there is little evidence to
suggest that Cornell is under-enrolled with regard to middle and upper
income students. While financial pressure on these populations is significant,
especially at the upper end of the aided group and the lower end of the no-
aid group, a large number of qualified students from these income groups
apply to Cornell, are admitted, and choose to enroll signaling their
willingness and ability to pay for a Cornell education. Far fewer students
from lower-income groups apply and are admitted, so it is important to
optimize enrollment from these more rare groups.

For each option presented, the AFAWG suggests a desirability rating of low,
moderate, or high. Financial implications are estimated based on the
assumption that each of the options in this report will be implemented; if
only some will be implemented, financial estimates of both costs and savings
should be revised.

Principles and Philosophy

Cornell University has long been committed to a philosophy and policy of

undergraduate need-blind admissions and undergraduate need-based
financial aid, including meeting full financial need as defined by institutional
methodology. This commitment upholds core values of fostering diversity
and academic excellence. Providing access based on competitive merit
requires allocating financial resources to enable all qualified and deserving
students to attend Cornell. Overall financial aid expenditures vary according
to the need of enrolled students; a defined budget limit is set for
international student grant aid. To more effectively manage limited grant aid
resources for international students, starting with students matriculating at
Cornell for Fall 17, international (and undocumented without DACA status)
students are considered in a need-aware admissions process. This change
means that all admitted students with need, whether US citizens, permanent
residents, or international, will have their full demonstrated need met
through financial aid (under the former policy, some admitted international
students with need received no financial aid).

The AFAWG agreed a primary goal for Cornells undergraduate financial aid
programs is to create as high-quality and diverse a community of scholars as
possible with the grant aid funds available. In practical terms, this means
that a significant portion (around 50% based on recent enrollment patterns
with need-blind admissions) of undergraduate students will not receive aid,
paying their full cost of attendance. Consequently, the other portion (~50%)
of undergraduate students will be aided, and should be distributed across
lower/middle/upper-middle socioeconomic communities, supporting diversity
of all types (e.g., race/ethnicity, geography, citizenship, disciplinary, gender,
political, economic, etc.).

The AFAWG endorses the concept that students and families should have a
significant stake in their Cornell education, reflecting their financial abilities
now and into the future. Parent contributions to the cost of attendance while

2
the student is enrolled are reasonable to expect for most income groups,
student contributions should be expected for all income groups, and loans
may be an appropriate component of most students financial aid packages,
particularly considering the value of a Cornell education in enabling students
to earn enough income to reasonably pay back loans.

For context, a loan of $23,264 (2015 mean loan at graduation for Cornell
undergraduates) being repaid through a non-income-based loan repayment
approach would require annual payments of $2,826 over a 10-year period
($28,264 total) or annual payments of $1,474 over a 25-year period ($36,838
total).

The AFAWG endorses principles expressed in the report of the 2012 Financial
Aid Task Force:

Fundamental to Cornells approach in calculating financial need to be

addressed through financial aid are the core beliefs that students should
shoulder significant responsibility for covering the cost of their education,
that families (parents) should be a primary source for helping students meet
educational costs, and that various types of financial aid allocated by Cornell
should be available for meeting the gap between the students potential
resources (including summer and academic year earnings, student savings,
parent contribution, loans, etc.) and educational expenses.

[These] principles reflect our core beliefs that our students benefit
immensely from studying with the most diverse and talented student body,
and our faculty benefit from teaching the most diverse and talented
students. Along with these fundamental values toward admissions and aid,
the university must also commit to sustainable fiduciary principles.

The AFAWG developed the following recommendations for consideration by

the university leadership, to inform choices that may need to be made
regarding responsible management of Cornells grant aid resources. Most of
these recommendations focus on possible controls on (reductions to) Cornell
grant aid expenditures. A few recommendations focus on updating cut-
points for certain financial aid policies to reflect changes in US income
quintiles since these policies were launched nearly 10 years ago, to address
concerns for students across socioeconomic groups; these changes would
result in modest increases to Cornell grant aid expenditures.

The following recommendations result from months of discussion and

analysis of a great deal of data including a comparison of Cornells financial
aid policies with those of Ivy and non-Ivy peers, student survey data
regarding factors affecting their Cornell enrollment decisions, Cornell student
debt and loan repayment trends, graduation rates for different aided and
unaided populations, enrolled student demographic trends and

3
socioeconomic diversity related to aid programs, and various income and
yield comparisons for aided and unaided populations. We also include
recommendations to keep certain current financial aid policies intact.

As noted in the Introduction, these recommendations provide options for the

university leadership to choose from, depending on university goals for grant
aid expenditures and considering the degree of control required on grant aid
expenditures. Having mechanisms available to control grant aid expenditures
is important should there be another economic downturn. In an anticipatory
sense to help Cornell be able to preserve its commitment to financial aid in
the event of another economic downturn, grant aid cost savings (expenditure
reductions) could be set aside in a reserve fund that would allow Cornell to
maintain a desired tuition discount rate in both economic good times as well
as economic downturn.

Based on modeling from Cornells budget office, any change in total

unrestricted grant aid cost that is equal to 1% of gross tuition will result in a
one-percentage point change in the discount rate. For example, FY15 gross
tuition revenue was $624M, unrestricted grant cost was $196M, and the
discount rate was 31.5%. Increasing unrestricted grant expenditures by
$6.24M would have increased the discount rate from 31.5% to 32.5%.
Similarly, decreasing unrestricted grant expenditures by $6.24M would have
decreased the discount rate to 30.5%. An estimator provides a general
guideline for interpreting the recommendations below: If Cornell leadership
seeks to achieve a 1% reduction in the discount rate, options that equal 1%
of gross tuition (as savings) could be selected from the list below.

To help Cornell leadership anticipate when additional controls (cost-savings

measures) may need to be implemented, recent modeling suggests that the
financial need population is correlated with unemployment rates and total
non-farm payroll (see figure below for actual need population vs. predicted
based on this measure). Thus, significant change in financial need trends
may be predictable such that the University could implement changes to
financial aid policies based on employment forecasts to avoid significant
unanticipated increases in financial aid expenditures as experienced
associated with the post-2008 economic downturn.

4
Recommendations

The AFAWG provides 8 recommendations below, some with multiple parts.

Most have specific savings or costs estimated for them, based on current
enrollment, but they vary in level of desirability as judged by the AFAWG
(low, moderate, high). If all recommendations are implemented, the total
net savings (savings minus costs) would be about $7.5M annually when fully
implemented across all four years of undergraduate enrollment. If all
recommendations are implemented, the percent of Cornell students
graduating with loans would increase from 42% (2016) to as much as 47%,
and the mean debt at graduation for those graduating with debt could
increase from $23,264 (2016) to about $24,500 (based on current enrollment
patterns).

1. Continue $0 parent contribution for students from families with <$60K

annual income and <$100K assets. Desirability: High
2. Continue $0 loans for students from families with <$45K annual
income, but change the no-loan cut-off from $60K annual family
income to $45K annual family income, increasing the maximum annual
loan for the $45K-$60K annual family income range from $0 to
$2,000/year. Estimated impact: 650 students; Savings = $1.3M.
Desirability: TBD
3. Change the income cut-offs for maximum loan levels, and change the
maximum loan levels:
a) Change the upper-bound of the $60K-$75K annual family income
category to $60K-$85K, and change the maximum loan limit,
raising the loan limit for the lower part of this range and lowering
the loan limit for the upper part of this new range.
1) Increase the maximum annual loan for the $60K-$75K
annual family income range from $2,500 to $3,500/year.

5
 Estimated impact: 290 students; Savings = $290K.
Desirability: TBD
2) Decrease the maximum annual loan for students from
families with $75K-$85K annual income from $5,000 to
$3,500/year. Estimated impact: 210 students; Cost =
$315K. Desirability: TBD (NOTE: Without 3.a.1,
the decrease would be from $5,000 to $2,500/year,
for an estimated cost = $525K.)
b) Change the lower and upper bounds of the $75K-$120K annual
family income category to $85K-$135K, and change the
maximum loan limit, raising the loan limit for the lower part of
this range and lowering the loan limit for the upper part of this
new range.
1) Increase the maximum annual loan for students from
families with $85K-$120K annual income from $5,000 to
$7,000. Estimated impact: 709 students; Savings =
$1.4M. Desirability: TBD
2) Decrease the maximum annual loan for students from
families with $120K-$135K annual income from $7,500 to
$7,000. Estimated impact: 300 students; Cost =
$150K. Desirability: TBD (NOTE: Without 3.b.1,
the decrease would be from $7,500 to $5,000, for an
estimated cost = $750K.)
c) Increase the maximum annual loan for students from families
with >$135K annual income who have need from $7,500 to
$9,000. Estimated impact: 1,258 students; Savings =
$1.9M. Desirability: TBD
4. Continue post-admittance preferential URM parent contribution
reduction for income groups $60K-$120K. Desirability: TBD
a) Include post-admittance preferential URM parent contribution
reduction (15%) for income groups $120K-$135K to correspond
to income band changes above. Estimated impact: 41
students; Cost = $259K. Desirability: TBD
5. Eliminate post-admittance URM parent contribution reduction for upper
income groups from selected states >$135K. Estimated impact:
107; Savings = $1.25M). Desirability: TBD
6. Increase the number of admitted no-aid international students using
need-aware admissions (without decreasing the international aid
budget, so aided international student admissions continue as-is).
Estimated impact: Additional 150 non-aided international
students would yield Savings = $2.03M. Desirability: TBD
7. Implement a need-aware admissions policy for transfer students.
Estimated Savings: (to be determined based on savings
desired). Desirability: TBD
8. Explore the feasibility of Cornell joining the Private 529 college savings
program https://www.privatecollege529.com/OFI529/ through which

6
 families can purchase tomorrows tuition at todays rates.
Estimated impact: (to be determined). Desirability: TBD
9. CONSIDER ADDING IN AFAWG: Modify Cornells institutional
methodology for calculating need vs. no-need to move more of the
current no need population into the need group.

Report Narrative
US Income Quintiles for Context
When developing recommendations, the AFAWG recognized that current
Cornell financial aid policies were implemented in 2008-2009 and revised
significantly following the 2012 Financial Aid Task Force Report. US income
quintiles have changed over this period. Current US income quintiles are:
o Lowest: $0 - $30,311
o Second: $30,312 - $55,376
o Third: $55,377 - $86,310
o Fourth: $86,311 - $133,525
o Fifth: >$133,526

National attention is high regarding the distribution of enrolled students at

elite universities across US income quintiles. There is, and will continue to
be, political pressures at state and national levels to limit or reduce student
loan debt, to make the cost of attendance affordable especially for low and
middle income families, and for elite institutions to actively contribute to
upward economic mobility of graduates.

A recent New York Times article on economic mobility reported that some
universities are quite economically segregated, though Cornell was rated
fairly well regarding the chance a poor student has to become a rich adult,
measured as the percent of students whose families were in the bottom fifth
US income quintile when they were enrolled as students who as graduates
moved in to the top fifth income quintile as adults (59% for Cornell). On this
measure, Cornell was 3rd in the Ivy League and 34th out of 2,137 colleges
nationally. On a measure of overall economic mobility reflecting both access
and outcomes representing the likelihood that a student after graduation
would move up two or more income quintiles, Cornell rated 1,395th out of
2,137 colleges nationally (but 1st among the Ivies). In addition, graduation
rates at Cornell (93%) are among the highest in the U.S. and Cornell
graduates earn median starting salaries ($63,000) after graduation above
the U.S. median household income ($53,657).

The AFAWG observes it will be important for Cornell to continue to pay

attention to both access and outcomes, implementing policies that will

7
enable Cornell to continue to have an economically diverse student
population.

Cornell Grant Aid Expenditures

The chart below provides financial context regarding why it is important to
consider potential expenditure controls on Cornell grant aid. The chart below
illustrates sources of need-based grant aid, AY1995 AY2017. The dotted
lines for CU Restricted and Unrestricted represent five years of a special
$35M/year withdrawal from Cornells endowment to cover financial aid
expenses. The shaded areas represent periods of US recessions identified by
the Bureau of Labor Statistics. The solid vertical lines identify substantial
changes to Cornell financial aid programs: Fall 02, introduction of
preferential packaging for several priority populations; Fall 08, first phase of
loan-reduction and parent contribution-reduction initiatives; Fall 13, revisions
to these initiatives.

Unrestricted dollars for grant aid have leveled out over the past few years,
following several years of sharp increase (and prior years of modest
increase), due in part to the Fall 13 financial aid program changes, changing
student demographics, and improved global economy.

Peer Comparisons
The AFAWG extensively reviewed peer institution financial aid policies for
comparison with Cornells policies, for Ivy and non-Ivy peers:

8
 o Most peers examined, except for two non-Ivy peers, use need-blind
admissions for US citizens and permanent residents; most peers (Ivy
and non-Ivy) use need-aware admissions for international students.
o No Ivy peers award merit aid (grant aid not based on need), but most
non-Ivy peers do.
o Most Ivy peers do not include loans as part of financial aid packages,
and for those that do the maximums are lower than Cornells (typically
$5000 or $5,500 compared to Cornells current maximum of $7,500),
and those that award loans do so only for families with incomes
>$100K (Cornells current no-loan cut-off is $60K).
o Most non-Ivy peers award loans as part of financial aid packages, but
maximums are lower than Cornells, mostly ranging from $3,000 -
$5,500 although one school has a maximum of $7,500 (the same as
Cornell). The starting family income level for including loans in
financial aid packages varies for non-Ivy peers from about $40K to
$80K, although a few non-Ivy peers may include loans for any income
group.

Although data on the percent of students in each US income quintile enrolled

at Cornell compared to peer schools is not available, data from federal IPEDS
based on undergraduates receiving Title IV financial aid (Pell, SEOG, Stafford
Loan, Work-study) suggests that Cornells percentages in different income
groups are similar to peers:

Title IV Aid Recipients as a Percentage of Fall First-time Enrollment, AY2013-14,

Selected Peers, by US Income Quintiles.

Total Pct Pct Not

Reciving Reciving
$0- $30- $48- $75- $110 Title IV Title IV
30k 48k 75k 110k + Aid Aid
Columbia University 4% 4% 6% 7% 14% 35% 65%
Cornell University 3% 6% 7% 8% 18% 43% 57%
Duke University 6% 5% 5% 8% 13% 37% 63%
Johns Hopkins University 6% 4% 7% 7% 24% 48% 52%
Northwestern University 7% 5% 7% 7% 16% 42% 58%
University of Chicago 3% 3% 6% 6% 14% 31% 69%
University of Notre Dame 4% 3% 5% 7% 22% 42% 58%
University of Pennsylvania 3% 5% 7% 9% 20% 45% 55%

Cornells Economic Diversity

A study by COFHE (Consortium on Financing Higher Education Assessing
Grant Aids Influence on College Choice) that compared matriculation
decisions made by applicants admitted to more than one COFHE school
examined the effect that differences in grant offers have on the students
final choice of which school to attend. The results suggested that lower-
income students are especially sensitive to variations in grant awards and

9
low-income students are more sensitive to institutional grant aid than are
higher-income students. The study also found that sensitivity to grant aid
was unrelated to race and ethnicity.

The AFAWG examined extensive data regarding Cornells socioeconomic

diversity and enrollment patterns. In summary regarding economic diversity:
o There is little evidence to suggest that Cornell is under-enrolled with
regard to middle and upper income students. As an example, Fall 2015
enrollment (which is a typical year, based on trends over the past decade
or more) is shown below, for students receiving need-based Cornell grant
aid vs. those not receiving grant aid, according to US income quintiles at
that time. These data suggest that upper income students are the
majority at Cornell (55% do not qualify for need-based grant aid), and of
the portion (45% overall) who receive grant aid, 53% of grant-aided
students are in the fourth and fifth US income quintiles, 17% of grant-
aided students are in the third quintile, 18% are in the second quintile,
and only 12% are in the lowest income quintile.

Examining trends in the financial aid need population in terms of the

percentage of Cornell tuition that is discounted, the table below
demonstrates that the percent of students receiving a tuition discount of
75% - 100% has declined over the past six years, and the percent of
students with 0% discount (full pay) has increased over this same period.
The percent of students in the upper middle income tiers inferred from
their financial aid status (and thus tuition discount) has remained fairly
steady, with <25% discount remaining at 4%, 25%-49% tuition discount
fluctuating over the six years at 6%-7%, and 50%-74% tuition discount
fluctuating at 8%-9%. These trends reflect the aided vs. no-aid enrollment
patterns over time; before the 2008 economic downturn, 56% of enrolled
undergraduates were full-pay and 44% received Cornell grant aid, but at the
height of the downturn, 49% were full-pay and 51% received grant aid. The
trend also reflects the declining portion of enrollment of the lowest-income
students; percent receiving Pell grants fell from a high of 18% in 2012 to
15% in 2016.

10
These tables above and the chart below lead the AFAWG to conclude:
o Over time, the enrollment of non-aided students has been increasing.
o Since 2008, the largest shifts in Cornell enrollment among aided
students in the five US income quintiles and unaided students have
been between the unaided category and the fourth and fifth US income
quintiles, with numbers of students shifting in and out of these groups
as the overall economy has shifted. Enrollment numbers of other US
income quintiles have remained relatively stable over time.
o All of these indicators suggest that attending to socioeconomic
diversity must remain a priority for Cornell through our financial aid
policies.

Net Cost to Families

As shown in the figure below, the median cost of a Cornell education after
grant aid, in inflation-adjusted dollars, is lower now than it was in 2008 for
aided students from all U.S. income quintiles except the highest.

11
Had Cornells financial aid policies been structured to hold the median cost
for all aided income groups, including the highest, to be lower than in 2008
(or the same now as in 2008), grant aid costs to the university would have
been much higher over time. For example, to bring the median net cost of
the aided highest US income quintile group down from the current (2017)
cost to the 2008 cost, $9.5M additional Cornell grant aid in 2017 would be
required as in the table below.
Median
Net Cost Count *
Yea Cou (2016 Delt Delta
Sector r nt Dollars) a ("Cost")
Contra
ct In- 20
State 08 230 $33,161
20 $4,1 $1,206,
17 294 $37,265 04 563
Contra
ct
Out-
of- 20
State 08 227 $38,263
20 $4,8 $2,189,
17 455 $43,075 12 678
Endow 20
ed 08 836 $40,733
20 116 $5,2 $6,156,
17 9 $46,000 67 937
$9,553,
Total 178

Loans and Excess Loans Above Financial Aid Package Awarded to

Student
Based on February 2017 data, students with loans, median loan amounts,
and median Expected Parent Contribution (EPC) are portrayed in the table
below by different income bands, corresponding to Cornells current loan-
reduction income thresholds ($60K, $75K, $120K, with additional categories
>$120K) to help visualize loan use among higher-income families.

12
Table: Number of Students with Loan, Median Loan, and Median Expected Parent
Contribution by Income Bands, February 2017 (Not Available= students with no
family income data because they did not apply for or were not evaluated for
financial aid.)
Income Range Number Median Median Expected Range of Median
of Loan Parent EPC as Percent of
Students Contribution (EPC) Income
$0-$60K 990 $4,680 $0 0
$60K-$75K 345 $3,100 $6,500 10.8% - 8.6%
$75K-$120K 1093 $5,000 $12,600 16.8% - 10.5%
$120K-$150K 713 $7,500 $22,200 18.5% - 14.8%
$150K-$200K 871 $7,500 $33,100 22.1% - 16.6%
$200K-$250K 369 $7,500 $46,850 23.4% - 18.7%
$250K-$300K 156 $7,500 $50,800 20.3% - 16.9%
$300K-$400K 107 $6,500 $71,450 23.8% - 17.9%
$400K-$500K 46 $6,500 $101,900 >cost of
attendance
$500K+ 26 $6,000 $158,100 >cost of
attendance
Not Available 261 $7,500 NA NA

The table above includes all loans for students with loans whether awarded
as part of the Cornell financial aid package, or incurred voluntarily by
families. In Fall 2016, of 4,154 domestic students receiving loan as part of a
need-based aid package, 1,203 (29%) received loan in excess of the amount
originally assigned as part of their Cornell aid package. The mean excess
loan was $8,602, but the distribution is heavily right-skewed and peaks in the
$2,000 $3,000 range. The median excess loan was $4,385 (see table and
figure below).

The predominant spike occurring at exactly $2,000 appears to be largely due

to families requesting additional unsubsidized federal loan to apply towards
the parent contribution for many students, this particular amount is the
remaining federal loan available after all possible subsidized federal loan has
been applied for (see table below).

Table: Excess Loan Above Financial Aid Package

Median
Percenta Excess
Excess Loan Count ge Loan
$1 $1,999 174 14% $1,000
$2,000 202 17% $2,000
$2,001
5,000 264 22% $3,500
$5,000
10,000 220 18% $6,875
10,000
20,000 182 15% $15,000
20,000
30,000 102 8% $23,407

13
$30,000+ 59 5% $38,800
Total 1,203 100% $4,385

To discern the reasons why students take out excess loan beyond the level
assigned in their financial aid package, we examined a random sample of 50
student records from the set taking extra loan. For 68% of the students in the
sample, it was clear from explanatory comments made by counselors or from
related data that the extra loan was being applied towards the family
contribution: this set included nearly all cases where the excess loan was
greater than $5,000. In 14% of the cases, the extra loan appeared to be in
response to unusual packaging circumstances, such as complications with
outside aid benefits. In 18% of the cases the reason was not stated and was
unclear.

Applications for budget increase requests from aided students seeking

additional loan indicate that most loan increases are requested for the
purpose of meeting housing costs and health insurance.

The table below displays the distribution of excess loan by income level. The
table uses the income thresholds used in Cornells loan policies, and
suggests that the rate of incurring excess loan is greater among those in the
$60k-$75k income range, but the median amount of extra loan for this group
is below most other groups, as the median amount appears to be increasing
with income. According to a COFHE report on student borrowing, about a
fifth of lower-income students at COFHE schools that assign no loans in their
financial aid packages still borrow, typically borrowing amounts between
$2,200 - $2,900 if they do borrow. Comparatively, the Cornell data suggest
that a higher percent of no-loan students may be incurring loans compared
to other COFHE no-loan schools, but the amount that lower-income Cornell
students are incurring in excess loan is on par with other COFHE schools.

14
Table: Excess Loan by Income Category

Income Number in Number with Percentage Median Excess

Total Loan Excess Loan Loan
Population
<$60k 782 222 28% $2,264
$60k - $75k 301 127 42% $3,250
$75k - $120k 1,184 348 29% $4,500
$120k+ 1,887 503 27% $7,500
Total 4,154 1,203 29% $4,385

Figure: Median Excess Loan by Income Range

Recommendation Details
The recommendations below should be considered in this context of peer
comparisons, Cornells values, accomplishments, and future goals regarding
a diverse undergraduate population, and actual student financial
circumstances beyond calculated financial need based on Cornells
institutional methodology for determining need.

Rationale for each recommendation, and associated concerns, are

summarized below.

1. Continue $0 parent contribution for students from families

with <$60K annual income and <$100K assets. Desirability:
High

Rationale: According to a 2016 COFHE (Consortium on Financing Higher

Education) report to university presidents, among high school seniors with
SAT (or PSAT) scores of 1250 or higher, only an estimated 18% have family
incomes below $75,000. Overall, the pool of high-ability lower-income
students represents only about 2.2% of a typical American high school class.

15
(Note: in the context of this report, SAT/ACT scores are used as a rough
proxy for high ability and high-achieving students, recognizing the
incompleteness of such a measure. In practice, Cornell uses a holistic
admissions process and does not make admissions decisions based solely on
test scores.)

Lower-income students apply to COFHE schools at lower rates than other

students. In part, this is due to the one-off problem: relatively isolated
high ability students are much less likely to apply to a selective institution
and lower-income students are more likely to be in this situation. Therefore,
Cornell should focus its grant aid strategically on reducing cost-related
barriers to attendance for the lowest income high achieving students, as this
is a very limited applicant pool for which there is strong competition among
all of our peer institutions.

Even under a policy of $0 parent contribution for this income level, students
and their families are contributing to their Cornell education; even at the
lowest income levels students do not receive a free ride as is sometimes
claimed. In addition, except for the highest income group, the net cost of
attendance is reasonably similar for all students receiving Cornell grant aid
regardless of the tuition rate endowed or contract (Fall 2016 Cornell data
below).

Concerns: None identified; AFAWG strongly supports #1.

2. Continue $0 loans for students from families with <$45K

annual income, but change the no-loan cut-off from $60K
annual family income to $45K annual family income, increasing
the maximum annual loan for the $45K-$60K income group
from $0 to $2,000/year. Estimated impact: 650 students;
Savings = $1.3M. Desirability: TBD

Rationale: See the explanation on #1 above as rationale for continuing some

no-loan for the lowest income group (<$45K). In addition, no-loan means

16
simply that Cornell does not assign loan as part of the awarded financial aid
package. However, as noted earlier in the Excess Loans section, nearly
one-third of all aided students incur educational loans in excess of the
amount originally assigned as part of their Cornell aid package (median
excess loan was $4,385), often to meet the expected family contribution.

The rationale for changing the no-loan cut-off to $45K annual family income
reflects non-Ivy peer comparisons that indicate at least a small number of
non-Ivy peers award loans beginning at about this income level. In addition,
Cornell students have very low loan default rates (1-2%), graduation rates at
Cornell (93%) are among the highest in the U.S., and Cornell graduates
(Class of 2015) earn mean ($62,980) and median starting salaries after
graduation ($63,000) above the U.S. median household income ($53,657).
Graduates who are employed by a government or not-for-profit organization,
including teachers, may be eligible for various federal student loan
forgiveness programs.

Concerns: Lowering the no-loan cut-point from $60K to $45K will increase the
percent of Cornells class graduating with student loan debt. Cornell has
made strides in the past decade reducing this percent while the US average
has been increasing.

AFAWG believes it is very important for Cornell to continue to publicize the

availability of a no-loan category to be competitive among our peers (Ivy and
non-Ivy) in recruiting and yielding high-achieving low income students given
the limited applicant pool nationally. Inserting even modest loans into
Cornells financial aid packages for students in the lowest income group
would make Cornell stand out visibly as compared to all of our selective
competitors who have $0 loans in their aid packages. The pool of top low-
income students (as measured by academic ability) is limited. Thus, it is
important that Cornell have a very high yield rate for them.

For Cornell, data on the relationship between SAT scores and family
economic status suggest the following:
SAT/ACT scores are weakly but positively correlated with income.
The difference in median SAT/ACT score between the non-aided group
and the lowest income quintile is a little more than 100 points.
For every $10,000 increase in income, SAT/ACT score increases, on
average, by about 3.9 points, although the relationship flattens above
$100,000.
Because the income and SAT/ACT distributions are both quite skewed,
there are relatively few admitted students who are both low-income
and high SAT/ACT, thus it is important to maximize yield for this group
of admitted students.

17
The chart below illustrates the large differences in high-SAT/ACT base counts
across income categories. Among 1,054 Fall 2016 admits in the highest
Cornell SAT/ACT quartile, 4% were from the lowest two US income quintiles.
There are 56 non-aided high-SAT/ACT admits for each high-SAT/ACT admit
from the lowest US income quintile.

(Blue = enrolling; Red = non-enrolling)

Given the relatively limited pool of high-achieving lower-income students (as

explained in #1 above), Cornell must remain competitive for recruiting these
students as applicants and yielding those who are admitted. Currently,
Cornells freshman regular decision yield rates for lower-income students are
good (61%-73% for the two lowest US income quintiles), and these rates
need to be maintained given the high competition for these students among
our peer institutions. It is not a problem that Cornell yield declines with
family income, because higher income students tend to have higher test
scores, and thus provide a larger, more robust applicant pool from which
Cornell (and peers) can recruit and admit qualified students.

18
Freshman Yield Rates for Regular Decision Admits, US Citizens and

Equivalents

As indicated by a variety of research studies, low-income families may be

averse to assuming loan debt because of cultural concerns about repayment
burdens; focused outreach on income-based repayment programs and
Cornells very low default rate will be necessary to avoid losing this group of
applicants and being able to recruit admitted students given the keen
competition among our peers for the limited number of high-achieving lower-
income students.

Cornell should continue efforts and invest resources in recruiting high-

achieving lower-income applicants to build the applicant pool as possible. In
addition, Cornell should invest in positions in Career Services that will work
directly with low income (especially Pell) students so that they receive
additional attention for the career preparation and job search process. If low
income students are asked to take loans they should be fully supported in
finding careers that will enable them to pay back the loans.

3. Change the income cut-offs for maximum loan levels, and

change the maximum loan levels:
a) Change the upper-bound of the $60K-$75K annual family
income category to $60K-$85K, and change the maximum
loan limit, raising the loan limit for the lower part of this
range and lowering the loan limit for the upper part of
this new range.
1) Increase the maximum annual loan for the $60K-
$75K annual family income range from $2,500 to

19
 $3,500/year. Estimated impact: 290 students;
Savings = $290K. Desirability: TBD
2) Decrease the maximum annual loan for students
from families with $75K-$85K annual income from
$5,000 to $3,500/year. Estimated impact: 210
students; Cost = $315K. Desirability: TBD (NOTE:
Without 3.a.1, the decrease would be from $5,000 to
$2,500/year, for an estimated cost = $525K.)
b) Change the lower and upper bounds of the $75K-$120K
annual family income category to $80K-$135K, and
change the maximum loan limit, raising the loan limit for
the lower part of this range and lowering the loan limit
for the upper part of this new range.
1) Increase the maximum annual loan for students
from families with $85K-$120K annual income from
$5,000 to $7,000. Estimated impact: 709 students;
Savings = $1.4M. Desirability: TBD
2) Decrease the maximum annual loan for students
from families with $120K-$135K annual income from
$7,500 to $7,000. Estimated impact: 300 students;
Cost = $150K. Desirability: TBD (NOTE: Without
3.b.1, the decrease would be from $7,500 to $5,000,
for an estimated cost = $750K.)
c) Increase the maximum annual loan for students from
families with >$135K annual income who have need from
$7,500 to $9,000. Estimated impact: 1,258 students;
Savings = $1.9M. Desirability: TBD

Rationale: Cut-points for maximum loan levels have not changed

substantially since they were implemented in 2008; maximum loan amounts
were increased following the 2012 Financial Aid Task Force
recommendations. US income quintiles have changed during that time
period, so resetting the cut-points may provide limited financial relief to
middle and upper-middle income families.

Increasing loans for all income groups above $45K distributes the increased
financial burden across all family types, excluding the lowest income. The
minimum family income for loans ($45K) is within the upper range of the
second US income quintile and reflects the minimum used by a small number
of non-Ivy peers.

Cornell loan repayment default rates are very low (1-2%), and Cornell alumni
secure jobs that enable them to pay back loans. For the Class of 2015, the
mean starting salary for new graduates was $62,980 and the median starting
salary was $63,000, compared to the 2014 median US household income of
$53,657.

20
In addition, a COFHE report on Student Debt What Seniors Report
suggests that nearly half of student borrowers expect their parents to help
them pay off their student loans, and this expectation is closely related to
parent income; students from higher-income families are more likely to
expect that their parents will help them pay off their student loans.

Concerns: Mean debt at graduation for Cornell graduates who borrow will
increase, as will the percent of students graduating with loans. While US
mean debt has been generally increasing (except for a decrease 2012-2013),
Cornell mean debt had been decreasing from 2006 until 2011, then began
rising again. Cornell mean debt, however, remains below national figures.
Mean debt at graduation for 2016 Cornell graduates who borrowed was
$23,264 while the 2014 national mean loan debt for all graduates who
borrowed was $28,950 and the 2014 national mean for graduates of private
nonprofit colleges was $32,600. Increasing loans for Cornell students in
these various income levels will bring Cornell figures closer to national
figures.

21
Increasing loans for upper-middle and upper income aided students may lead
to decreased yield for these students. Although the applicant pools of high-
achieving upper-middle and upper income students are quite a bit larger and
more robust than applicant pools of high-achieving lower-income students,
nonetheless Cornell must remain reasonably competitive for these students
against peers, some of whom award merit aid (not based on need). For this
reason, the AFAWG offers a separate recommendation (#4) to reduce loan
for selected high-achieving higher-income students.

In addition, a practical but unknown consideration in the current national

higher education context is whether the federal Perkins loan program will
continue or be cancelled. If the Perkins program ceases to exist, it is
possible that an increase in loan caps could push students into the private
lending market unless Cornells own university loan program could substitute
for the defunct Perkins loan program.

4. Continue post-admittance preferential URM parent

contribution reduction for income groups $60K-$120K.
Desirability: TBD
a) Include post-admittance preferential URM parent
contribution reduction (15%) for income groups $120K-
$135K to correspond to income band changes above.
Estimated impact: 41 students; Cost = $259K.
Desirability: TBD

Rationale: Cornell has made marked gains in enrolling students from

underrepresented minority (URM) populations. Cornell and all of our peer
institutions compete actively to recruit and yield students from the
somewhat limited pool of high-achieving, middle income URM applicants, so
Cornell should maintain its ability for effective yields for this population.

22
Several schools are particularly strong competitors against Cornell for
yielding Black and Hispanic admitted students and have comparatively
generous financial aid programs for these students at all income levels, so at
least for the middle income groups Cornell should strive to make a Cornell
education financially attractive for these students, and can do that by
retaining this preferential URM treatment.

Cornell Undergraduate Enrollment, by Ethnicity, 2006-2016

For example, in Fall16, the top competitors with Cornell for yielding admitted
Black students (525 total admits for Cornell) were University of Pennsylvania
(yielded 27 Cornell admits), MIT (23), Harvard (21), Princeton (21), Duke (15),
Stanford (14), Yale (11), Brown (10), Johns Hopkins (9), University of Chicago
(9). This resulted in Cornell enrolling 228 Black admitted students, while
these top competitors enrolled 160 of Cornell-admitted Black students (and

23
137 Cornell-admitted black students enrolled at other schools not in this top
list). Competition for these same students is fierce among elite universities,
and Cornell must continue to be competitive in this pursuit.

Hispanic non-enrolls (from among 989 Cornell admits) were similar in terms
of top peer competitors: MIT (24), Brown (23), Harvard (23), Stanford (22),
Yale (22), University of Pennsylvania (21), Johns Hopkins (19), Duke (17), UC
Berkeley (17), Princeton (16), but not as skewed to a relatively small set of
key competitors. Of 989 Cornell-admitted Hispanic students, 461 enrolled at
Cornell, 214 enrolled at the top competitors, and 314 enrolled at a variety of
other institutions.

Across all admits, for every $5,000 decrease in net cost, the yield rate
increases by about 3 percentage points. This has been consistent over the
past seven years. URM students (of any income) generally yield at a lower
rate than non-URM students but still exhibit about a 3% increase in URM
yield for every $5,000 decrease in net cost. Continuing to limit the URM
parent contribution for this income band will help Cornell be able to continue
enrolling a racially/ethnically diverse undergraduate population.

Additionally, upward URM enrollment trends primarily reflect Hispanic

population increases. African American enrollment trends nationwide appear
to be declining, suggesting that if Cornell seeks to maintain or grow our
African American student population, competition with our peers for high-
achieving African American students is likely to intensify.

Concerns: Eliminating the post-admittance preferential URM parent

contribution reduction for income groups $60K-$120K would
produce $2.73M in Cornell grant aid cost-savings, affecting ~1,250
students. This is a substantial cost savings that could meaningfully
contribute to Cornells efforts to limit total grant aid expenditures.

Preferential URM parent contribution reductions are not publicly announced

and thus have no impact on the decision to apply to Cornell, because URM
families dont know that this is a possible benefit they might receive if
admitted to Cornell. Cornells demonstrated success in increasing URM
enrollment at Cornell coupled with national demographic trends suggest
there may be continuing upward URM enrollment trends even without special
Cornell grant aid (parent contribution reduction) treatment. In addition,
because this is not advertised publicly, removing or reducing the post-
admittance preferential (decreased) URM parent contribution could be
treated as an experiment that could be reversed rather quickly if needed, if
URM enrollment of these income groups showed a sharp decline seemingly
related to a change in this behind-the-scenes financial aid policy.

24
A COFHE report on Assessing Grant Aids Influence on College Choice found
that although lower-income students may be more sensitive to differences
in aid awards than those with higher incomes, this choice elasticity is
unrelated to race and ethnicity, calling into question the basic assumption
on which this Cornell policy is grounded.

5. Eliminate post-admittance URM parent contribution reduction

for upper income groups from selected states >$135K.
Estimated impact: 107 students; Savings = $1.25M.
Desirability: TBD

Rationale: The 2012 Financial Aid Task Force Report recommendations led to
Cornell implementing several financial aid program changes that had been
targeted specifically toward URM students. Starting with Fall 13 enrollees,
Cornell eliminated preferential loan reduction for URM students, saving about
$5.7M of grant aid. In addition, Cornell changed from using a single level of
URM parent contribution reduction to a graduated parent contribution-
reduction algorithm based on family income.

A COFHE report on Assessing Grant Aids Influence on College Choice found

that although lower-income students may be more sensitive to differences
in aid awards than those with higher incomes, this choice elasticity is
unrelated to race and ethnicity, calling into question the basic assumption
on which this Cornell policy is grounded and providing rationale for this
Recommendation.

Making these income-related changes to the special URM parent contribution

reduction had no discernible negative impact on Cornells ability to recruit
and yield URM students. When pre- and post-2013 samples are balanced
based on various factors (SAT/ACT scores, recruiting region, college of
admission, gender, URM status, etc.), the resulting impact of the financial aid
program changes on yield is nearly non-existent. In the context of concerns
regarding changes in post-admission preferential financial aid treatment for
URM students, we have not observed substantive reductions in yield of
admitted students, nor reduction in applications. (Of course, it is impossible
to know what additional gains may have been achieved in the absence of the
changes implemented in Fall 13.)

URM (U.S.) Fall Freshmen Yield: Any student who identifies as one or more of the
following is classified as URM (U.S.): American Indian (U.S.), Black (U.S.), Hispanic (U.S.),
and/or Hawaiian/Pac Isle (U.S.).
College Year Apps Admits Deposits Admit Rate Yield
CORNELL 2011 6,301 1,464 673 23.2% 46.0%
2012 6,258 1,479 670 23.6% 45.3%
2013 6,623 1,518 724 22.9% 47.7%

25
 2014 7,478 1,547 707 20.7% 45.7%
2015 7,595 1,615 756 21.3% 46.8%
2016 8,161 1,723 790 21.1% 45.9%

In Fall 2016 3,118 URM undergraduates were enrolled at Cornell. The 107
students affected by this change represent 3.4% of total URM undergraduate
enrollment.

A Spring 15 analysis presented to the Board of Trustees Task Force on

Faculty, Staff, and Student Diversity concluded as follows:
Financial aid program changes implemented in Fall 2013 do not appear to
have had an adverse effect on undergraduate diversity.
URM undergraduate enrollment is increasing 6%/year, whereas non-URM
enrollment is fairly stable or decreasing slightly.
o Enrollment of unaided URM students has increased 10%/year.
o Enrollment of aided URM students has increased 5%/year.
o Enrollment of unaided non-URM students has increased 1%/year.
o Enrollment of aided non-URM students has decreased 2%/year.
o URM first-generation enrollment has increased 4%/year, whereas
non-URM first-generation has decreased 3%/year.

In addition, this financial aid program element can be changed nearly

instantaneously if we observe detrimental effects. In other words, because
this program element is not advertised publicly and is known only to
students after they have been admitted to Cornell and awarded their
financial aid packages, Cornell would be able to reinstitute this parent
contribution reduction very easily should there be evidence that changing it
leads to a significant reduction in URM student enrollment.

Concerns: Similar to the rationale for #5 (which was to continue the URM
parent contribution reduction program for income groups <$120K and
extend it to groups <$135K), by implementing this change Cornell runs the
risk of losing higher-income URM students to top competitor institutions with
more generous financial aid packages. URM student yield is lower, at all
income levels, compared to non-URM student yield.

26
Cornell should be concerned that it is easier to recruit a significant number of
new URM students when they see on campus an existing critical mass of
URM students. If URM student enrollment begins to decline, it may keep
declining.

6. Increase the number of admitted no-aid international students

using need-aware admissions (without decreasing the
international aid budget, so aided international student
admissions continue as-is). Estimated impact: Additional 150
non-aided international students would yield Savings =
$2.03M. Desirability: TBD

Rationale: This option could be potentially turned up or down fairly quickly

with more significant impact than other options and with potentially less
negative public perception issues than other options that are more visible. In
addition, larger numbers of international students contribute to Cornells
undergraduate diversity. No-aid international students may be truly self-
pay (from higher-income families), or Cornell could focus greater resources
on recruiting relationships with countries that provide full scholarships to
their students to be able to attend US universities (not necessarily higher-
income families).

Using this approach could help Cornell return to or remain at its prior
historical levels of percent of students receiving grant aid (in the low 40s)
vs. percent of students who are unaided (upper 50s).

27
Concerns: As with recent discussions when Cornell changed from need-blind
international student admissions to need-aware admissions there will be
concerns voiced that this option would inappropriately privilege wealthy
international students for admittance to Cornell.

7. Implement a need-aware admissions policy for transfer

students. Estimated Savings: (to be determined based on
savings desired). Desirability: TBD

Rationale: A fixed budget for transfer enrollment, similar to the existing use
of a fixed budget for international students, would improve Cornells ability to
more actively control financial aid expenditures. Implementing this option
would not necessarily lead to a reduction of transfer students, nor
necessarily lower-income transfer students. Rather, using need-aware
admissions would help Cornell make explicit, strategic choices about how to
spend its grant aid dollars on transfer students. Cornell could affirmatively
decide to fully fund admitted Transfer Option students as part of a need-
aware policy. The financial aid budget for transfer students could be set at
the expenditure level that currently exists to provide financial aid to all
enrolled transfer students with need. Setting such a realistic limit based on
current enrollment could help prevent significant increases in financial aid
expenditures should another serious economic downturn occur.

For the university overall, and for each college/school, the demographic
profile of the enrolling transfer class is generally less diverse than the
freshman class, although on measures of economic diversity, transfer

28
students in some colleges/schools are more likely than freshmen to be
eligible for need-based financial aid in some years. Socioeconomic diversity
among transfers varies considerably from year to year. Using a need-aware
admissions approach could help stabilize economic diversity (aided v.
unaided) among transfer enrollment from year to year.

Transfer students who enter Cornell from community colleges would

presumably be entering with relatively low debt, so could reasonably be
expected to pay for, through loans or other means, the rest of their Cornell
education especially given the very high graduation rates at Cornell and the
relatively high post-graduation salaries Cornell graduates achieve.

At least one Ivy peer, Brown University, uses need-aware admissions for
transfer students.

Concerns: Some colleges (especially CALS and ILR currently) are more
dependent on transfer students than others. Transfer students represent a
relatively small portion of the overall undergraduate student body,
approximately 12-15% of new student fall enrollment each year. The portion
differs by college/school, however, with significant portions of the graduating
classes of ILR (40-45%), CALS (30%) and SHA (25%) comprising transfer
students. If implemented, care would need to be taken to limit or manage
the impact for colleges/schools in which transfer students are important
contributors to goals for student enrollment.

In addition, implementing need-aware admissions for transfer students could

lead to negative effects on the various articulation agreements that Cornell
undergraduate colleges (CALS, CHE) have with community colleges.

8. Explore the feasibility of Cornell joining the Private 529 college

savings program https://www.privatecollege529.com/OFI529/
through which families can purchase tomorrows tuition at
todays rates. Estimated impact: (to be determined).
Desirability: TBD

Rationale: Nearly 300 schools currently participate in the Private College 529
program, including MIT, Princeton, Stanford, Duke, Chicago. The Private
College 529 Plan is most likely to benefit those who can already devote some
financial resources to their childrens education, and want to make their
money go further. It is likely that groups of middle-upper income families
would benefit the most. The Plan appears largely compatible with Cornells
current financial aid policies and packaging approaches.

A COFHE report on COFHE Parents and College Costs suggested that the
specific financing plan for a students education affects the familys
perceptions of impact on the family. The study found that college financing

29
elements that are associated with lower negative impacts include employee
benefits, pre-paid tuition and 529 plans (such as the Private 529 program),
savings, and gifts from relatives.

Cornells participation would support our message of doing what we can to

keep the cost of tuition affordable for all families. The Plan should be
received positively by families with disposable income to contribute to the
fund, and could assist families with long-range financial planning for college.

Because the fund is an asset of the account owner (typically parent or

grandparent), it may mean that when implemented in our institutional
financial aid methodology it would show as an increase in parent
contribution, thus reducing Cornell grant aid.

Concerns: To determine the financial cost/revenue implications for Cornell,

the universitys budget and finance offices are undertaking an analysis that
will inform further consideration of this option. Based on data from the
Private 529 program, it is unlikely this would include a large number of
Cornell families, as the entire program nationwide has about 6,000
participants and close to 300 schools enrolled.

Cornell would have to be careful with communications about the Private 529
Plan. Unlike the public 529 plan where saved funds can be applied to the full
cost of education, the Private 529 Plan can only be applied to the cost of
tuition. We would have to be careful in our communications so that families
do not expect an advantage in the admissions decision process.

9. CONSIDER ADDING IN AFAWG: Modify Cornells institutional

methodology for calculating need vs. no-need to move
more of the current no need population into the need
group.
Add analysis started below.
Although the number of non-aided students enrolled has been generally
increasing since 2010 (see figure below), the number of non-aided students
incurring educational loans since that time has decreased from 356 to 261,
as has the number of non-aided students incurring loans <$30K/year
(decrease from 306 to 181). The number of non-aided students incurring
annual loans >$30K increased somewhat, from 50 to 80.

30
31
Appendix: Additional Charts and Figures

Tuition Discount Trends

Cornell tuition discount trends should be interpreted considering context:

In FY08 and FY09, Cornell adopted and implemented generous changes
to financial aid programs, including no-loan, no-parent contribution,
and loan cap policies for various student populations.
In Fall 13 (FY14), Cornell implemented changes to financial aid
programs based on recommendations from the 2012 Financial Aid Task
Force that, when fully implemented, would result in $21M annual
savings.
Fall 16 (FY17) was the first year in which the 2012 Financial Aid Task
Force report recommendations were fully implemented across all 4
undergraduate cohorts. The total discount rate decreased over that
time period from 42% to 35%.

Year % Unrestricted % Total

Discount Discount
FY08 19.2% 27%
FY09 25.0% 33%
FY10 23.8% 38%
FY11 27.8% 41%
FY12 27.3% 42%
FY13 28.8% 42%
FY14 26.2% 39%
FY15 31.5% 38%
FY16 29.8% 37%
FY17 29.3% 35%

Distribution of Student Loans by Income Band

All income groups peak at about $7K-$10K and then drop to a long tail. The
NA group is students who did not apply for or were not evaluated for financial
aid.

32
Black and Hispanic Enrollment Trends

33
Incoming Freshman Self-Reports (via Survey) of Family
Income, 2008-2016
Since 2000, Cornell has surveyed its incoming class of first-year students.
Most years include a question asking students to report the household
income of their families. Below are data from the 2008, 2010, 2012, 2014,
and 2016 surveys. Response rates ranged from 83% to 88% depending on
year. Confidential survey responses were merged with administrative data
on student financial aid status for the 2008 2014 survey years.

These data do not adjust for inflation; the proportion of students coming from
families earning >$250K has grown from a fifth or less to more than a
quarter of the student population (left panel). The profile of the unaided
student group (center panel) has shifted from one that is 36% highest
income (19% out of 52%) to one that is close to half highest income (24%
out of 49%). Some shift up is to be expected due to inflation.

34
Graduating Seniors Self-reported Impact on Family
Finances, 2006-2016

Below are data from the 2006, 2010, 2012, 2014, and 2016 surveys of graduating
seniors. Response rates ranged from 45% - 50% depending on year. Confidential
survey responses were merged with administrative data on student financial aid
status. The survey question was What has been the impact on your family of
paying for your education (tuition, fees, rooms and board) at Cornell? where
responses ranged from slight to severe. The percent of seniors reporting the impact
of paying for college is severe or considerable decreased from 50% in 2006 to
47% in 2016 (left panel).

35
The table below shows the percent of graduating seniors reporting that the
impact has been considerable or severe. Darker red means the impact was
worse. The length of the bar is the number of students in that income/aid
category. Among those most inclined to say the impact was severe or
considerable are aided students with family incomes between $50K and $200K.

A COFHE report on COFHE Parents and College Costs suggested that the
specific financing plan for a students education affects the familys
perceptions of impact on the family. The study found that financing
elements associated with higher negative impacts include federal PLUS

36
loans, a second job, student borrowing, retirement account borrowing or
withdrawals, mortgage, private loans, and real assets.

Cornell Pell Grant Percent Compared to Peers

Fall 2014
Fall 2014 Pell Pct of
Pell Count Enrollment
University of Southern California 4,323 23%
Columbia University 1,777 22%
Massachusetts Institute of
Technology 806 18%
California Institute of Technology 37 16%
Stanford University 1,104 16%
Brown University 1,047 16%
Princeton University 790 15%
Rice University 607 15%
Cornell University 2,207 15%
Dartmouth College 596 14%
Duke University 903 14%
Northwestern University 1,245 14%
University of Pennsylvania 1,607 14%
Vanderbilt University 956 14%
Georgetown University 957 13%
Yale University 705 13%
Harvard University 1,204 12%
Johns Hopkins University 781 12%
Tufts University 620 12%
University of Chicago 639 11%
University of Notre Dame 888 11%
Washington University in St Louis 548 7%

Study Abroad/Off Campus Program Participation by

Income
These are the rates of CU Abroad/Off Campus program participation by aid
status, for AY15 and AY16. Aided is defined as students receiving need-
based grant aid from Cornell sources. It appears that low income students
are most likely to participate in Abroad or Off-campus programs, in both
years. Aided students in roughly the fourth and fifth US income quintiles
(reflected in Cornells $120-$200K group) appear least likely to participate in
Abroad or Off-campus programs, in both years.

37
 $75- $120- Aided Not
Fall $0-75k 120k 200k Total Aided

2014-
15 5.7% 4.6% 4.3% 5.0% 4.8%

2015-
16 4.9% 4.2% 3.8% 4.3% 4.6%

For the table above, the chart below provides the counts by income/aid status for
the two years (based on students enrolled in the Fall of each year):

Distributions of Aided and Unaided Populations

38
Percentage of Enrolling Freshmen with Pell or Other Need-based Aid

Income of Undergraduates Receiving Need-based Cornell Grant,

FY16, URM vs. Non-URM

39
SAT and INCOME
SAT total scores are the sum of the Critical Reading and Mathematics
sections, which (for the years being studied) can produce scores ranging
from 200 to 800 in 10-point increments (so a total SAT score can range
from 400 to 1600). Cornell does not include the SAT writing section is not
included in the SAT total score uses in most reporting. The ACT exam
(which is in theory designed to measure different mental constructs from the
SAT) produces scores that can range from 1 to 36, in one-point increments.
For much internal and external reporting at Cornell, ACT scores are converted
to the SAT scale using a score concordance jointly prepared by the College
Board and the ACT organization. Of 6,337 students admitted to Fall 2016,
about 73% submitted SAT scores and about 50% submitted ACT scores, and
about 20% submitted both SAT and ACT scores.

In the table below are SAT/ACT-equivalent quartile limits used in this report,
which are based on admitted students from both Fall 2015 and Fall 2016. For
this reason, these figures may differ trivially from figures used in other
reports. The SAT/ACT distribution among students admitted to Cornell is also
quite asymmetrical, skewed toward high scores (see histogram).

40
Note that the College Board reports that the SAT standard error of
measurement is approximately 30-40 points e.g., differences smaller than
40 are not necessarily inconsistent with random noise.
SAT/ACT Score Summaries for 2015 and 2016 Admitted Freshmen
Income Quintile/Aid Status
Secon Highes
Total Lowest Third Fourth No Aid
d t
Median SAT/ACT 1480 1380 1420 1440 1460 1490 1500
75th Pctl 1550 1460 1510 1520 1540 1550 1550
25th Pctl 1400 1270 1300 1340 1370 1400 1420
Max 1600 1600 1600 1600 1600 1600 1600
Min 870 920 990 990 950 980 870
Enrolling Median 1460 1360 1380 1420 1430 1460 1460
Non-Enrolling
1510 1430 1460 1485 1510 1510 1510
Median

Relationship between Income and SAT/ACT-equivalent

Among aided students, those with higher incomes tend to have slightly higher
SAT/ACT scores than those with lower incomes, but the relationship is very weak.
For first time freshmen admitted for Fall 2015 and Fall 2016, the R-squared from
regressing income on SAT/ACT-equivalent (with no additional predictors) was 0.052.
The regression coefficient for income indicates that for every $10,000 increase in
income, SAT/ACT score increases, on average, by about 3.9 points. Across the
income range from $0 to $300,000, this equates to about a 100-point difference.
However, the relationship is not quite linear. The R-squared from a GAM model
(curvy regression using additive splines) was only slightly higher, but suggested
that the relationship between income and SAT/ACT is slightly stronger at the lower
end of the income range, flattening out above about $100,000 in income.

Strength of Relationship between SAT/Act and Income

Fall Linear R-sq Income Coefficient * GAM R-sq
$10,000
Total 0.052 3.9 Points 0.069
2015 0.051 4.0 points 0.065
2016 0.053 3.7 points 0.076

41
In the charts below, each small blue dot represents an enrolling admit, and
red dots indicate non-enrolling admits. A small amount of random spread has
been added to points at each categorical axis point to help in visualizing the
distributions.

This chart shows the distributions of SAT/ACT scores for admitted freshmen
from Fall 2015 and Fall 2016 combined, with median SAT/ACT scores overlaid.
Counts and yield rates (combined across years) are displayed across the
bottom.

Relationship between Yield, SAT/ACT, and US Income Quintiles

Admitted students with higher SAT/ACT score tend to yield at lower

rates.
For the two years being analyzed, a 100-point increase in SAT/ACT
score was associated with a roughly 13 percentage point decrease in
the likelihood of yield.
Although aided students are about 25 percentage points more likely to
enroll than unaided students, the relationship between SAT/ACT and
likelihood of yield among aided students is fairly consistent for both
aided and unaided students.
Similarly, although average yield rates vary across income and aid
categories, the basic relationship between SAT/ACT and yield is fairly
consistent across groups.

42
All results in this section are based on domestic, regular decision first-time
freshmen admitted to Fall 2015 or Fall 2016. The table below summarizes
admit and enrollment counts, yield rates, and median SAT/ACT scores for
each income quintile and SAT/ACT quartile. Yield rates range from a low of
28% for non-aided, high-SAT/ACT admits, to 83% for low-income, low-
SAT/ACT admits.

SAT/ACT
Quartile Lowest Second Third Fourth Highest No Aid
Highest Admitted 38 77 119 242 446 1,957
Enrolling 22 30 59 106 167 545
Yield 58% 39% 50% 44% 37% 28%
Median
SAT/ACT 1560 1560 1560 1570 1570 1570
Third Admitted 60 143 135 248 453 2,370
Enrolling 40 76 87 129 207 1,028
Yield 67% 53% 64% 52% 46% 43%
Median
SAT/ACT 1510 1510 1510 1510 1510 1510
Second Admitted 120 176 160 274 434 2,102
Enrolling 93 119 106 172 263 1,188
Yield 78% 68% 66% 63% 61% 57%
Median
SAT/ACT 1450 1460 1450 1450 1460 1450
Lowest Admitted 259 333 276 367 459 1,563
Enrolling 215 258 222 286 325 923
Yield 83% 77% 80% 78% 71% 59%
Median
SAT/ACT 1280 1300 1315 1320 1340 1340

43