in debt and in the dark: it’s time for better information on student loan defaults

7/28/2019 In Debt and In the Dark: Its Time for Better Information on Student Loan Defaults

1/13

www.educationsector.org

charts you can trust

In Debt and In the Dark:Its Time or Better Inormationon Student Loan Deaults

By Andrew Gillen

Few domestic issues resonate more deeply with the public than the

continually rising cost of college. With soaring tuitions (and oten disappointing

completion rates) have come demands or cheaper alternatives and stricter

institutional accountability. The responsibility or both has traditionally resided

with colleges, but the ederal government can inuence postsecondary behavior

in two key ways that relate to student loan deault rates: First, by providing better

inormation on student loan deault rates. Second, by holding colleges more

accountable or their part in student loan deaults.

The three-year deault rate or student loans is 13.4 percent, and the cost odeault will be borne by studentsand taxpayers.1

Tracking and reporting loan deault rates are a crucial means o monitoring how

well higher education dollars are spent; it is a primary ederal responsibility.

Better loan deault data would hold institutions to a stricter level o accountability,

provide more useul inormation to students and parents, and help researchers

determine which students struggle most to aord college, why they struggle, and

how to address those problems. Yet, despite the debates about ways the ederal

government might improve accountability and reduce student debtor example,

by providing meaningul data on industries where graduates fnd jobs, by trackinggraduation rates or part-time students, or by increasing oversight o college

accreditorspolicymakers have not devoted enough attention to overhauling one

basic step: how the ederal government measures and reports data on student

loan deaults.

Incomplete Loan Deault Data

Data on loan deaults is reported by institution. That means that whether you are

looking at an elite Ivy or a local community college, you can see the percentage

o borrowers who attended those schools and deaulted at some point in the

frst three years o the repayment period.2 But thats all you can learn. There are

no breakouts o that data. Consider Pell grants, given primarily to poor students.

I you want to see how many o one schools Pell grant students deaulted

compared to how many o the schools other students deaulted, you are out o

luck. Likewise, you cant fnd out how deaults among Pell grant students at one

school compare to deaults o Pell grant students at another school. You also cant

look at deault rates or dierent types o degree programsbachelors versus
http://www.educationsector.org/http://www.educationsector.org/


2/13

www.educationsector.org July 2013


masters versus doctoral at a our-year university, or associate versus certifcate

programs at a community college. Nor can you look at deault rates by students

age, gender, or ethnic group.

None o this inormation is available despite the act that deault rates are one

o the primary means by which the ederal government determines eligibility orederal aid. (Schools also must be accredited and approved in their home states,

but the ederal government does not directly inuence these decisions).

Given the importance o deaults, and the recent jump in their numbers, it would

make sense or the government to provide more detailed inormation on deaults,

not just as an accountability lever but as a basic consumer right. Compared to

other orms o debt, it is remarkably easy to qualiy or a ederal student loan

(though remarkably difcult to discharge in bankruptcy): Good credit histories are

not required or most loans, and the risk to students is shared by taxpayers since

the ederal government is the direct lender or ederal loans. Parents and students

would beneft rom better data showing which students, at which schools, are

more likely to deault on higher education loans.

Elsewhere in education, actions by the ederal government have ooded

policymakers, parents, and other taxpayers with important and useul data. The

ederal No Child Let Behind Act remains controversial, but because o it, at the

K12 level, results o achievement tests are broken down by state and district.

Results also are available by race and categories such as English profciency and

special education status, so that schoolwide averages no longer mask dierent

levels o perormance among sub-populations.

Why cant the same rigor be applied to reporting requirements or deault rates

at the college level? Already the ederal government has made some changes in

how it calculates deault rates. Colleges will soon be judged on the percentage o

What is a Student Loan Default?

When student borrowers graduate, drop out, or all below hal-time enrollment,

its time to start paying back their student loanswhat the government calls

entering repayment. Most ederal loans have a six-month grace period

beore entering repayment. Student borrowers also can be granted deerment

(a temporary reprieve rom loan payments) or orbearance (an alternative

payment schedule) to postpone or alter scheduled payments.

The U.S. Department o Education considers a loan to be in deault i the

borrower is more than 270 days behind on payments. The deault rate is the

percentage o a schools borrowers who enter repayment during a scal

year and deault within three years. Note: Only subsidized and unsubsidized

Staord loans are included in the deault rate calculation. The deault rate

calculation ignores Parent PLUS, Grad PLUS, and Perkins loans.


3/13


charts you can trust3

borrowers who deault in the frst three years o the repayment period rather than

the percentage who deault in the frst two years. (Three-year deault rates were

published or the frst time in 2012). Under the new rules, schools with deault rates

greater than 30 percent or three consecutive years will lose access to ederal

student aid. (Lesser consequences kick in beore that.) Under the two-year rule,

the eligibility threshold was 25 percent.3

Two-Year vs. Three-Year Rates

Congress instituted the three-year rule because o concerns that schools were

gaming the two-year requirement by delaying the date at which students would go

into deault (through orbearance, deerment, and other means). Since this would

allow colleges to avoid accountability or their students deaults, taking action

to prevent such schemes seemed a reasonable response. But Chart 1, which

shows the two-year and the three-year deault rate or each college, indicates that

with ew exceptions, i you know a schools two-year deault rate, you can pretty

accurately predict its three-year deault rate. Colleges largely did not hide deault

rates by shiting them outside the old two-year measurement window.

There were a ew outliers such as Lassen Community College in Caliornia. Its

two-year deault rate was 15.2 percent; its three-year deault rate jumped to 37.7

percent. Several or-proft colleges also saw unusually large jumps. For example,

at ITT Technical Institute, which has many campuses, deault rates or most o the

system rose rom 18.2 percent to 34.1 percent.

Deault Rates as an Accountability Tool

Using deault rates as an accountability mechanism is appealing or two reasons:

First, the rates provide an objective and quantifable measure o a colleges

success in providing a cost-eective education. For many students, the main

reason or enrolling in college is to improve their lietime fnancial well-being,

and a high deault rate indicates that a college has ailed to help its students

achieve this important goal.

Second, deault rates are relatively hard or colleges to game. They are a

better accountability tool than, or instance, graduation rates, which a college

can improve by lowering its standards. To avoid punishment or high deault

rates, a college could lower tuition (reducing the amount students need to

borrow) or provide a better education, which should improve the job prospects

o its graduates and their ability to repay their loans.


4/13



There are two drawbacks to using deault rates as an accountability tool:

First, employment outcomes, even when calculated on a value-added basis,

are not a complete measure o educational outcomes. As hundreds o

academics have argued, there is a lot more to being college-educated than

getting a good job. What this means in practice is that deault rates should not

be the onlyaccountability tool used.

Second, some students are more likely than others to deault or reasons

that have nothing to do with the cost or the quality o their education. Auent

parents oten can help their children with loan payments, or instance, while

low-income parents oten cannot. Thereore, we would expect the deault

rate to be higher at a college that educates more low-income students than

at a college that educates more high-income students, even i the colleges

are otherwise identical. An accountability system that ailed to account or

HART 1

Two-year andhree-year

deault ratesmatch closely.

Two-Year Default Rate

Three-YearDefaultRate

0 10 20 30 40

0

10

20

30

40

50

60

Source: Integrated Postsecondary Education Data System, Student Financial Aid, and authors calculations.

1) Each dot represents a school a.) with sufcient data, b.) or which at least 100 students entered repayment inFY2009, and c.) with at least 250 ull-time-equivalent students in 2009 10.

2) The red line indicates the best ft linear regression line.

3) On average, a colleges three-year deault rate was 1.46 times its two-year deault rate. More than 92 percent othe variation in three-year deault rates is explained by variation in two-year deault rates.


5/13



this dierence would punish colleges or serving students who are at greater

fnancial risk.

Fortunately, it is possible to devise input-adjusted or predicted deault rates to

create a better accountability system, one that makes allowances or dierences in

students backgrounds. Under this system, external actors that inuence deaultrates would be taken into account, adjusting a colleges actual deault rate or

actors such as amily income. While we cannot know all the inuencing actors,

analysis with available data illustrates the wisdom o accounting or probable

deault-risk actors.

The frst step is to select the risk actors. Income is one. Unortunately, there is

almost no publicly available data on the income o students at each college. But

we do know what percentage o students receive Pell grants, and since these

grants are awarded largely to students rom low-income amilies, the percentage

o Pell grant recipients can serve as a good proxy or the proportion o low-income

Finding the Predicted Default Rate

To nd the predicted deault rate, Education Sector ran a regression analysis

using the deault rate as the dependent variable, and the percentages o

undergraduates receiving Pell grants and part-time undergraduates as the

independent variables.

Four-Year Colleges Regression Results

Estimate Std. Error t value Pr(>|t|)

Constant -2.134891 0.256028 -8.339 < 0.001***

% Pell 0.284719 0.005995 47.49 < 0.001***

% Part time 0.037259 0.006677 5.58 < 0.001***

N = 1,600, Adj. R2 = .61

Two-Year Colleges Regression Results

Estimate Std. Error t value Pr(>|t|)

Constant 13.181325 1.106419 11.913 < 0.001***

% Pell 0.125689 0.014634 8.589 < 0.001***

% Part time 0.009817 0.012049 0.815 0.415

N = 1,052, Adj. R2 = .11

The coecient estimates were combined with each colleges actual values

or the percentage o Pell grant recipients and percentage o students who

were part time to yield the predicted deault rate. For example, a our-year

college with 30 percent o students receiving Pell grants and 10 percent o

students attending part time would have a predicted deault rate o: -2.13489

+ 30 * 0.284719 + 10 * 0.037259 = 6.8%. This predicted rate can be compared

to the colleges ocial rate to determine i the college is doing better or

worse than expected.


6/13



HART 2

Four-yearcolleges with

similar studentshave widelyvarying deaultates.

students at a college. Similarly, the percentage o part-time students can serve

as a proxy or nontraditional students, a category that also may present a higher

deault risk.

These two variables (percentage Pell and percentage part time) were used to

calculate each colleges expected or predicted deault rate. Charts 2 and 3 plotthe predicted deault rate against each colleges actual deault rate.

I a colleges actual deault rate is lower than its predicted deault rate, its dot

is below the red line in Charts 2 and 3, and that college is doing better than

expectedgiven the students it educates. I a colleges actual deault rate is higher

than its predicted deault rate, its dot is above the red line in Charts 2 and 3, and

that college is doing worse than expected. It is likely that ewer o these students

would deault i they attended other schools. Rather than holding colleges


1) Each dot represents a our-year college a.) with su fcient data, b.) or which at least 100 students entered

repayment in FY2009, and c.) with at least 250 ull-time-equivalent students in 2009 10.

2) The red line shows all points or which the predicted deault rate is equal to the actual deault rate.

3) Predicted values are each colleges expected deault rate, based on a regression using as independent variables

the percentage o students who are Pell grant recipients and the percentage o students who are par t time.

Predicted vs. Actual Student Loan Default Rates: 4-Year Colleges

Predicted Default Rates Based on % of Pell and Part-Time Students

ActualDefaultRate

0 10 155 20 25 30

0

10

20

30

40


7/13



accountable or whether they meet hard caps (such as an unadjusted deault rate

below 30 percent), we should hold colleges accountable or whether they are

signifcantly above or below the red predicted-deault line.4 While this example only

accounts or two inputs (because other desired data was not available), it does

show the wisdom o designing accountability thresholds that are input-adjusted.

Consider two public, our-year universitiesCentral State University in Ohio and

Southern University at New Orleans. Seventy-seven percent o students at Central

State receive Pell grants and 7 percent attend part time. Seventy-six percent o

students at Southern University at New Orleans receive Pell grants and 21 percent

attend part time.

HART 3

Two-yearcolleges with

similar studentshave widervariationn deault rateshan our-year

schools.


1) Each dot represents a two year college a.) with sufcient data, b.) or which at least 100 students entered

repayment in FY2009, and c.) with at least 250 ull-time equivalent students in 2009 10.

2) The red line shows all points or which the predicted deault rate is equal to the actual deault rate.

3) Predicted values are each colleges expected deault rate, based on a regression using as independent variablesthe percentage o students who are Pell grant recipients and the percentage o students who are par t time.

4) There are several colleges (e.g. ITT Technical Institute) that report systemwide deault rates, but campus-specifc student characteristics. This results in slightly dierent predicted values (based on variations in studentcharacteristics by campus), and the same actual values (based on the systemwide deault rate), which appear on

the graph as a group o horizontally arrayed dots.

2016 18 22 24 26 28

0

10

20

30

40

50

Predicted Default Rates Based on % of Pell and Part-Time Students

ActualDefaultRate


8/13



Statistical analysis indicates that the typical deault rate or colleges that have

similar student characteristics is around 20 percent. But Central State Universitys

deault rate is 32.8 percent; Southern University at New Orleans is 17 percent.

Students rom Central State are deaulting at a much higher than expected rate,

while students rom Southern are deaulting at a lower than expected rate.

This example shows how just two inputs can improve accountability metrics. I

the ederal government releases data on other appropriate input variables, these

calculations would provide a uller picture, one that better shows which schools

are more successul at preparing students or lie ater college.

Pairing Deault and Graduation Rates

The 1990 Student Right-To-Know Act required colleges to report their graduation

rates or frst-time, ull-time degree- or certifcate-seeking students. Deault rates

are also a key piece o consumer inormation or college students. Chart 4 shows

that graduation rates plus deault rates provide a clearer understanding o each

schools outcomes.

For example, Gettysburg College (Pennsylvania), St. Ola College (Minnesota), and

Martin Methodist College (Tennessee) are all private, nonproft, our-year-and-above

colleges with similar Carnegie Classifcations. All have an impressive graduation

rate o 85 percent. However, Gettysburgs deault rate is 0.4 percent, and St. Olas

is 1.6 percent. Yet Martin Methodists is 26.6 percent. (Keep in mind however,

that ewer o Martin Methodists students borrow.) Students are well-advised to

consider more than simply the graduation rate when choosing a college.

Toward a Better Understanding o Deaults

The ederal government has long played a vital role in education research. But too

oten, the voluminous data the U.S. Department o Education collects ocuses on

inputs (such as stafng levels, expenditures, and incoming student test scores)

and stops well short o providing the kind o data that could provide answers to

important public policy questions. This shortcoming applies especially to higher

education, where many colleges remain stubbornly resistant to providing detailed

inormation around outcomes, and the government has not required them to do

otherwise.

The U.S. Department o Education is now the sole originator o ederal student

loans. Given the soaring costs o college, the still-ragile job market, and rising

rates o college loan deaults, the ederal government should oer better data or

analysis and research to reveal which types o students, in what felds and degree

programs, are most likely to deault. The importance o this inormation is clear.


9/13



Yet, despite lending more than $100 billion annually, the government requires

minimal standards or students to qualiy and leaves taxpayers on the hook or

billions o dollars in deaults.

This needs to change.

Until we have better data on loan deaults, the ederal government will continue

to lend billions to students every year with little to show students, taxpayers, or

policymakers about what happens when those students have to pay back that

money. Recent rule changes will let us know i these borrowers deault within the

frst three years o repayment, but beyond that, each borrowers status will remain

a mystery. Further, when deault rates improve as we emerge rom recession, they

HART 4

Deault Rates+ Graduation

Rates = BetterDecisions.


1) Each dot represents a school a) with sufcient data, b) or which at least 100 students entered repayment inFY2009, and c) with at least 250 ull-time equivalent students in 2009 10.

2) The red line indicates the best ft linear regression line.

3) Graduation rates are the Student-Right-to-Know rates or 2009 10.

4) There are several colleges (e.g. ITT Technical Institute) that report systemwide deault rates, but campus-specifcgraduation rates, which appear on the graph as a group o horizontally arrayed dots.

Graduation Rate

Three-YearDefaultRate

0 20 40 60 80 100

0

10

20

30

40

50

60

70


10/13



may well mask shortcomings in certain degree programs that, even in a strong

economy, represent poor career investments or students.

At the same time, colleges that do a superior job o educating Pell students and

have lower than expected deault rates will still have higher deault rates than other

schools that enroll relatively ew Pell students. Better data would highlight thesecolleges accomplishments.

We have seen how changes in ederal policy have led to markedly better outcomes

data at the K12 level. The ederal government spends even more money on

higher education, and postsecondary students assume more risk. With stakes this

high, students and taxpayers need to know ar more about what they are getting

themselves into. Improving deault-rate data would be a good place to start.

Red Flags: When Default Rates TopGraduation Rates

Exasperated fnancial aid expert Tim Ranzetta proposed that

some colleges post this disclosure:

WARNING: This education can be hazardous to your fnancial

health. At this institution, you have a higher probability o deaulting

on your student loan than you do o completing this program.5

These colleges should set o a red ag in the minds o prospective student

borrowersand their parents. Many students at these colleges will no doubt

take out loans, graduate, and get good jobs. But the high deault rates and

lower graduation rates suggest that many students will not.

Comparing graduation rates and deault rates or the same students would

produce a list o these red ag colleges. But the U.S. Department o Education

does not have that list, and the data it releases doesnt allow others to easily

put one together. Instead, graduation rates are tracked only or frst-time,

ull-time degree- or certifcate-seeking students (ignoring part-time, returning,and transer students); deault rates are tracked or students based on the

time period when they enter repayment.6 In addition, campuses with several

branches (such as the University o Phoenix) oten have one systemwide deault

rate and branch-by-branch graduation rates.

Nevertheless, using the ofcial graduation and deault rates as the best

estimate o the overall graduation and deault rates or college borrowers, we


11/13



can identiy the schools that are the most likely candidates to set o red ags.

These are colleges where students who borrow are more likely to deault on

those loans than they are to graduate. For example, New River Community

and Technical College in West Virginia has a graduation rate o 5 percent and a

deault rate o 25.7 percent.7 It is not alone.

O the 514 red ag colleges identifed, 314 (61 percent) are public, two-year

colleges. While these schools tend to have low tuitions, they oten serve

at-risk students. Among our-year colleges, the list included 100 or-profts, 48

nonprofts, and 29 public colleges.

Red Flag Colleges

OfcialDeault Rate >

OfcialGraduation

Ratea

OfcialDeault Rate >

OfcialGraduation Rate

& at least30% o Students

Borrowb

OfcialDeault Rate >

OfcialGraduation Rate

& at least100 Students

Deaultedc

AdjustedDeault Rate >

OfcialGraduation

Rated

Sector

Private or-proft2-year

19 19 16 12

Private or-proft4-year or above

100 97 96 64

Private nonproft4-year or above

48 36 33 18

Public2-year 314 88 188 18

Public4-year orabove

29 22 24 6

Other 4 3 2 2

Total 514 265 359 120


Notes: This table includes every college with a) sufcient data, b) or which at least 100 students enteredrepayment in FY2009, and c) with at least 250 ull-time equivalent students in 2009 10. There are several

colleges that report systemwide deault rates, but campus-specifc student characteristics. Each campus witha lower graduation rate than the systemwide deault rate was counted as a separate college. We also excludedroughly 100 colleges whose Ofce o Postsecondary Education identifcation number (OPEID) could not be

matched to an OPEID number in the deault rate database.

a) The ofcial graduation rate is or 2009 10. The ofcial deault rate is the three-year rate or all students whoentered repayment between Oct. 1, 2008 and Sept. 30, 2009.

b) The ofcial graduation rate is or 2009 10. The ofcial deault rate is the three-year rate or all students who

entered repayment between Oct. 1, 2008 and Sept. 30, 2009. The percentage o students who borrow is defnedas the percent o undergraduate students who took out a ederal loan in 2009 10.

c) The ofcial graduation rate is or 2009 10. The ofcial deault rate is the three-year rate or all students who

entered repayment between Oct. 1, 2008 and Sept. 30, 2009. The number o students who deaulted is rom theStudent Financial Aid database.

d) The adjusted deault rate is the colleges three-year deault rate or all students who entered repayment

between Oct. 1, 2008 and Sept. 30, 2009 multiplied by the percentage o undergraduate students who took outederal loans in 2009 10.

ABLE 1

Hundreds o

colleges shouldset o red fagsor borrowers.


12/13



Among these colleges, there is considerable variation in the percentage o

students borrowing, so the second column in Table 1 shows only those

colleges where at least 30 percent o undergraduate students took out ederal

loans. The number o colleges drops to 265, o which 97 are or-proft, our-year

colleges; 36 are nonproft, our-year colleges; 88 are public, two-year colleges;and 22 are public, our-year colleges.

The third column in Table 1 shows only those colleges where the absolute

number o students who deaulted is at least 100. The result is a new list o

359 colleges. Most o the change in the list occurs among 188 public,

two-year colleges.

The last column o Table 1 provides a fnal alternative list. Multiplying the

deault rate by the percentage o undergraduate students who took out

ederal loans yields an adjusted deault rate. This new rate estimates the

prevalence o deault among the whole student body rather than just among

borrowers. For example, Los Angeles City College, Blue Ridge Community

and Technical College (West Virginia), and Lane Community College (Oregon)

are all public, two-year community colleges with an ofcial deault rate o 19.5

percent, indicating that borrowers at all three schools have a similar likelihood

o deaulting. But there is considerable variation in the percentage o students

who borrow. Only 4 percent o Los Angeles City Colleges students took out a

ederal loan in 2009 10, but 33 percent o Blue Ridge students and 65 percent

o Lane students did. Using this inormation to calculate an adjusted deault rate

that applies to all studentsas opposed to just borrowersestimates that lessthan 1 percent o all students at Los Angeles City College deaulted on their

student loans, 6.4 percent o Blue Ridge students deaulted, and 12.7 percent

o Lane Community Colleges students deaulted.

O the 120 red ag colleges using the adjusted rate, more than hal (64) are

our-year, or profts. There are 18 public community colleges and 6 public,

our-year universities.

Table 1 makes two main points. First, some student loan borrowers are

attending colleges where it appears that they are more likely to deault than

they are to receive a degree. Second, red ags will pop up across all types o

schoolspublic, private nonproft, and private or-proft.


13/13


www.educationsector.org

ABOUT THE AUTHOR

Andrew Gillen is the research director at Education Sector. He can be reached [email protected] .

ACKNOWLEDGMENTS

Id like to thank all those who oered invaluable eedback. Susan Headden and Carol Knopes spent

many hours helping improve this report. Any remaining errors or omissions are my own.

ABOUT EDUCATION SECTOR

Education Sector is an independent think tank that challenges conventional thinking in education policy.

We are a nonproft, nonpartisan organization committed to achieving measurable impact in education, both

by improving existing reorm initiatives and by developing new, innovative solutions to our nations most

pressing education problems.

Notes

1. U.S. Department o Education, Three-year Ofcial Cohort Deault

Rates or Schools, available at http://www2.ed.gov/ofces/OSFAP/

deaultmanagement/cdr.html

2. Prior to 2012, the two-year deault rate was reported.

3. Details on the sanctions tied to deault rates can be ound at

http://iap.ed.gov/DeaultManagement/guide/attachments/

CDRGuideCh2Pt4CDREects.pd

4. The line used or accountabi lity purposes should take into account the

confdence intervals or the predicted deault rates.

5. Tim Ranzetta, Warning: This Education Could Be Hazardous To Your

Financial Health, Student Lending Analytics Blog, October 18, 2009. http://

studentlendinganalytics.typepad.com/student_lending_analytics/2009/10/warning-this-education-can-be-hazardous-to-your-fnancial-health.html

6. The graduation rate used in this study is the Student Right-To-Know rate or

2009 10. The three-year deault rates used in this study are or those students

who entered repayment in fscal year 2009 (between Oct. 1, 2008 and Sept.

30, 2009).

7. This graduation rate is rom the U.S. Department o Educations Integrated

Postsecondary Education Data System (IPEDS). The deault rate is rom the

U.S. Department o Educations Student Financial Aid database.
http://www.educationsector.org/mailto:agillen%40educationsector.org?subject=http://www2.ed.gov/offices/OSFAP/defaultmanagement/cdr.htmlhttp://www2.ed.gov/offices/OSFAP/defaultmanagement/cdr.htmlhttp://ifap.ed.gov/DefaultManagement/guide/attachments/CDRGuideCh2Pt4CDREffects.pdfhttp://ifap.ed.gov/DefaultManagement/guide/attachments/CDRGuideCh2Pt4CDREffects.pdfhttp://studentlendinganalytics.typepad.com/student_lending_analytics/2009/10/warning-this-education-can-be-hazardous-to-your-financial-health.htmlhttp://studentlendinganalytics.typepad.com/student_lending_analytics/2009/10/warning-this-education-can-be-hazardous-to-your-financial-health.htmlhttp://studentlendinganalytics.typepad.com/student_lending_analytics/2009/10/warning-this-education-can-be-hazardous-to-your-financial-health.htmlhttp://studentlendinganalytics.typepad.com/student_lending_analytics/2009/10/warning-this-education-can-be-hazardous-to-your-financial-health.htmlhttp://studentlendinganalytics.typepad.com/student_lending_analytics/2009/10/warning-this-education-can-be-hazardous-to-your-financial-health.htmlhttp://studentlendinganalytics.typepad.com/student_lending_analytics/2009/10/warning-this-education-can-be-hazardous-to-your-financial-health.htmlhttp://ifap.ed.gov/DefaultManagement/guide/attachments/CDRGuideCh2Pt4CDREffects.pdfhttp://ifap.ed.gov/DefaultManagement/guide/attachments/CDRGuideCh2Pt4CDREffects.pdfhttp://www2.ed.gov/offices/OSFAP/defaultmanagement/cdr.htmlhttp://www2.ed.gov/offices/OSFAP/defaultmanagement/cdr.htmlmailto:agillen%40educationsector.org?subject=http://www.educationsector.org/

in debt and in the dark: it’s time for better information on student loan defaults

Documents