Abstract October 2020 - JPMorgan Chase
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Student Loan Debt: Diana Farrell
Fiona Greig
Who is Paying it Down? Daniel M. Sullivan
October 2020
Abstract
American families carry more than spouses—in order to manage their and 7 percent of all borrowers not
$1.5 trillion in student loan debt. This debt. In particular, we find that while in deferral are on track to never pay
debt provided many with the oppor- the median borrower is not unduly off their loans. These dynamics of
tunity to pursue higher education, but burdened by their debt, a significant repayment put Black borrowers at a
remains for others a large, potentially minority of lower-income and younger disadvantage, who, relative to White
crippling, financial burden. In this borrowers are heavily burdened, borrowers, have lower incomes and
report, we explore how people of required to make payments that con- higher debt balances and are 4 times
different socioeconomic groups are stitute more than 10 percent of their as likely to have no payments made
managing their student debt. We do take-home income. We also find that against their loans, partly due to the
this by linking administrative banking almost 40 percent of those involved fact that they are less likely to receive
data, credit bureau records, and public in student debt repayment are making repayment help. This debt provided
records on race and ethnicity to create payments on other people’s loans, many with the opportunity to pursue
a unique data asset that includes the with 27 percent of those involved higher education with commensurate
income, demographics, debt balances, holding no student debt whatsoever. income keeping debt burdens at
and student loan payments of 301,583 These outside helpers play a key role reasonable rates. For others, student
individuals. In general, we find that in helping borrowers make progress loan debt remains a large financial
borrowers of socioeconomic groups on their loan. Nevertheless, we find burden relative to income. In this
tend to manage student loans quite that low-income and older borrow- report, we explore how people of
differently, often relying heavily ers are more likely to be several different socioeconomic groups
on others—children, parents, and months behind on their payments, are managing their student debt.
About the Institute
The JPMorgan Chase Institute is harnessing the scale and scope of one of the world’s leading firms to explain the global
economy as it truly exists. Drawing on JPMorgan Chase’s unique proprietary data, expertise, and market access, the Institute
develops analyses and insights on the inner workings of the economy, frames critical problems, and convenes stakeholders
and leading thinkers.
The mission of the JPMorgan Chase Institute is to help decision makers—policymakers, businesses, and nonprofit leaders—
appreciate the scale, granularity, diversity, and interconnectedness of the global economic system and use timely data and
thoughtful analysis to make more informed decisions that advance prosperity for all.Table of Contents
3 Executive Summary 23 Finding Four
Compared to White and Hispanic student loan
borrowers, Black borrowers are less likely
7 Introduction to be making progress on their loans.
12 Finding One 29 Implications
Although the median student loan borrower
is obligated to pay 3.8 percent of their take-
home income, many borrowers, especially 32 Data Asset
lower-income and younger borrowers, face
payment burdens well over 10 percent.
35 Appendix
15 Finding Two
Almost 40 percent of individuals involved in
36 References
student loan repayment are helping someone else
pay off their student loan debt, with most helpers 38 Endnotes
holding no student loan debt themselves.
40 Acknowledgements and
19 Finding Three
Suggested Citation
Low-income and older borrowers are more
likely to be behind on payments or in
deferral, and roughly 7 percent of borrowers
are projected not to repay their loans.Executive Summary
American families carry more than $1.5 trillion in student loan debt. This debt provided many with the opportunity to pursue
higher education with commensurate income keeping debt burdens at reasonable rates. For others, student loan debt remains
a large financial burden relative to income. In this report, we explore how people of different socioeconomic groups are
managing their student debt.
Finding One Median reported payment burden by income level
10%
Reported payment as a
percentage of income
8%
Although the median student
loan borrower is obligated to 6%
pay 3.8 percent of their take-
4%
home income, many borrowers,
especially lower-income and 2%
younger borrowers, face payment
0%
burdens well over 10 percent. 20 30 40 50 60 70 80 90 100 200
Income ($1,000s)
Note: Medians are calculated wit in twenty income quantiles. Eac income bin is represented by its average
income. Reported payment is total amount paid toward outstanding student debt during t e twelve-mont
window December 2015 t roug November 2016. Income refers to take- ome income.
Source: JPMorgan Chase Institute
Finding Two Individuals who do not have a student Pure Helpers These individuals ( 9 percent)
loan but have made payments (No student loan debt are helping someone else
towards student loans. 27% pay down their student loan
debt by making student loan
Almost 40 percent of individuals payments towards loans that
Individuals who have a student loan and
involved in student loan repayment have made payments but whose Net Helpers are not theirs.
12%
are helping someone else pay payments are also helping pay down
another person’s student loan.
off their student loan debt,
with most helpers holding no Individuals who have a student loan and Some of these
student loan debt themselves. have made student loan payments Paying Debtors individuals might be
out of their checking account 43% receiving help from others to
but are not Net Helpers. the extent that their reported
payments exceed their
Individuals who have a student loan but observed payments.
have not made payments towards
Non-Paying Debtors
student loans out of their
18%
checking account.
Source: JPMorgan Chase Institute
Student Loan Debt: Who is Paying it Down? Executive Summary 3Finding Three
Low-income and older borrowers are more likely to be behind on payments or in deferral, and roughly 7 percent
of borrowers are projected not to repay their loans.
Distribution of payment shortfall by age Distribution of payment shortfall by income
6 10 percent of borrowers with incomes less than 6
Payment shortfall in months
$30,000 in take-home income are 4 to 6 months
Payment shortfall in months
5 or more behind on their payments in just one year. 5 10 percent of borrowers around age 60 are
at least 3 months behind in their payments.
4 4
90th Percentile
3 3
2 2
The median (50th percentile) borrower
1 90th Percentile 1 around age 60 is current with payments.
0 50th Percentile 0 50th Percentile
20 40 60 80 100 200 30 40 50 60 70
Age Income ($1,000s)
Note: Percentiles are calculated within twenty income and age quantiles, respectively. Each bin is represented along the x-axis by its average value. Payment shortfall is
the difference between all scheduled and reported payments during the twelve-month window December 2015 through November 2016, divided by average monthly
scheduled payment. Income refers to take-home income.
Source: JPMorgan Chase Institute
Finding Four
Compared to White and Hispanic student loan borrowers, Black borrowers are less likely to be making progress on
their loans.
Progress on student debt repayment by race
13 percent of Black borrowers not
in deferment are on track to never
Nearly 10 percent of Black pay off their student loans in that
borrowers had no payments their loan balance is increasing.
made against their student
loans.
13.1%
9.9%
8.4%
6.8%
4.5%
2.6%
Black Hispanic White Black Hispanic White
No payments made against loan On track to never pay off
Note: The sample is restricted to borrowers who do not have a student loan in deferral or forbearance during the twelve-month window December 2015 through November
2016. Borrowers projected to never pay off debt have increasing balances over the twelve-month sample period; that is, interest charges over the course of the year are
larger than total payments made. Income refers to take-home income.
Source: JPMorgan Chase Institute
4 Executive Summary Student Loan Debt: Who is Paying it Down?In summary, this report finds that without saying that curbing the rise step in expanding targeted assistance
student debt holders are not a in tuition costs and student loan debt would be to help additional borrowers
monolithic group. Many borrowers borne by students and their families benefit from improved access to
are not unreasonably burdened would address the problem at its existing payment assistance programs,
by student loan payments and are root. In addition, reducing racial gaps such as IDR. Student loan debt policies
making payments on time. But in income and wealth would boost and assistance programs should also
certain segments of the student loan families’ ability to pay for tuition take into consideration the extent
population are significantly burdened and repay student loan debt among to which students rely on a network
by their debt, especially low-income segments of the population most of people to repay their student
borrowers, the elderly, and Black burdened by student loan debt. loans. Loan origination programs
borrowers. Moreover, we find that a Setting aside these structural issues might want to rebalance eligibility of
significant portion of student debt that contribute to the patterns of loans between students and parents.
payments are made not by the loan student loan repayment that we Additionally, there could be more
holder, but by other individuals not observe, we explore a few possibilities avenues for payment assistance for
tied to the loan, presumably family for how targeted debt assistance parents. A possible complement to
members who may not directly reap programs could be expanded to repayment relief programs is to allow
the labor market returns to higher alleviate the burden of existing student for restructuring or forgiveness of
human capital investment. This means loan borrowers. As a general principle, student debt through a bankruptcy-like
that the economic impacts of student because the majority of borrowers process.1 A further step to address
debt likely affect a broader portion are managing their debt without undue payment burdens would be to
of the population than previously being excessively burdened, efforts expand efforts to provide targeted
thought. Additionally, the prominent to alleviate undue burdens from debt forgiveness to those most
role of help in student loan repayment student loan debt should be targeted burdened. Targeted student loan
puts Black borrowers at a disadvantage at those who are facing truly difficult debt forgiveness could be a means
in that they exhibit a greater unmet circumstances. This is true for payment of rebalancing our investments in
need for repayment assistance. assistance efforts like income-driven public goods such as education across
What should be done to address the repayment (IDR) programs as well communities and insuring against
disparate patterns we find in student as more aggressive actions like debt the risk that borrowers, Black and
loan borrower outcomes? It goes forgiveness. A relatively easy first Hispanic borrowers disproportionately,
find themselves in a debt trap.
Student Loan Debt: Who is Paying it Down? Executive Summary 5Data Asset
We assembled a novel dataset of Louisiana—this also includes self- we consider an account in active use
301,583 de-identified Chase checking reported race and ethnicity data taken if it has at least five transactions in
account customers who had out- from public voter registration records. every month of our sample period
standing student debt or were making We constructed our sample of 301,583 and at least $12,000 in deposits over
payments towards student debt. from a larger match of 4.75 million the course of the sample period. This
We linked Experian credit bureau Chase customers to Experian records gave us a base sample of 1.8 million
data for December 2015 through covering December 2015 through customers. From these, we selected
November 2016 to these individuals’ November 2016. We then restricted all individuals who either (a) have an
bank data. This joint data asset this sample to those customers who open student loan in the Experian
allows us to observe income, student meet certain activity criteria in order records or (b) make payments out
loan payments, and key attributes to ensure a reliable analytical sample. of their Chase account to a student
of the student loan tradeline (e.g., Customers in our sample must have loan servicer, leaving us with our
origination date) and the account been Chase customers for the entire final sample of 301,583 customers
holder (e.g., age). For three states period of study. They must have also involved in student debt repayment.
in our sample—Florida, Georgia, and actively used their Chase accounts;
Universe of 39 million Chase checking accounts
Sample of 4.75 million Experian records who have Chase checking accounts
1.8 million “core” Chase checking accounts with Experian Records
(have $12,000 of deposits and five transactions per month)
301,583 Chase checking accounts who are involved in student loan repayment
(either hold student debt or have made at least one payment to a student loan ser vicer)
80,873 People making student loan
220,710 Student loan holders
payments but not holding a loan
16,799 People involved in student loan repayment for whom we observe self-reported race
(from 2018 voter registration files in Florida, Georgia, and Louisiana)
6 Executive Summary Student Loan Debt: Who is Paying it Down?Introduction
Student debt allows more people to officially scheduled to make on their ameliorate student debt burden ought
attend college with the prospects loan. Measuring the distribution of to consider not only the borrower
of achieving higher income levels, actual and scheduled payment burden but also the network of people the
but many policymakers and student is especially important in light of the borrower relies upon, who notably
advocates argue that the burden income-driven repayment programs, may not benefit financially from
of the debt—totaling over $1.5 and additional measures taken in the the human capital investment.
trillion—presents a looming crisis COVID-19 crisis, which attempt to align Third, who is making progress on
(Looney and Yannelis 2015). With the student loan repayment obligations repaying their student loan debt?
cost of higher education rising far with borrowers’ ability to pay. An important metric of success for
faster than inflation, an increasing student loan debt is the ability of the
number of students and their debtor to pay off their loan over time.
families use student loan debt to Evidence suggests, however, that a
finance their education. For many, Understanding number of borrowers are stuck in a
this is a sound investment paid off how families share the
“debt trap,” with student loan balances
through higher earnings over time. burden of student debt is
increasing rather than decreasing
For others, however, unpaid student important for the design of
over time even as they attempt to pay
loans become a lifelong burden with both loan origination and
repayment programs. them down (Gibbs 2017). What share
minimal returns. What impact does of borrowers are in this situation?
student debt have on the financial How much time does the typical
lives of borrowers and their families? borrower take to repay their loan?
In this report, we develop a new data Finally, are there racial disparities
asset that links credit bureau records Second, who shoulders the burden in student loan debt and repayment
with administrative banking data from of student loan debt during patterns? Survey evidence suggests
Chase checking accounts in 2016. repayment? With a growing share that there are large racial disparities
Together, these two data sources of federal student lending for in student loan borrowing and
provide information on outstanding undergraduates now composed of repayment, showing that Black
student debt, total monthly payments Parent PLUS loans, student loan debt individuals are more likely to take on
made against the debt, income, and is increasingly shouldered by both student loan debt and experience more
household demographics for 1.8 the recipients of higher education difficult conditions for repayment.3
million families.2 Using this dataset, and their family members (Baum et We contribute to this literature by
we answer four key questions. al. 2019). In this report, we are able assembling an administrative data
First, what is the payment-to-income to distinguish between payers who set that pairs banking data with
burden of student loan debt? In are legally tied to the student loan voter registration data in order
our recent report, we noted that one debt versus those who are making to expose disparities in payment
in four families spend more than payments on another person’s burdens, progress, and help received
11 percent of their take-home income behalf. This allows us to shed light in repaying student loan debt across
on student loans in months with on the extent to which student loan White, Black, and Hispanic families.
positive payments (Farrell et al. 2019). repayment is unofficially a “family
affair.” Understanding how families While the economics literature has
Here we take this analysis one step documented the impact of student
further, measuring burden by not only share the burden of student debt
is important for the design of both debt on many variables, from college
using payments made out of the loan- completion to home ownership, we
holder’s checking account, but also loan origination and repayment
programs insofar as policies to lack a comprehensive understanding
in terms of the payments they were
Student Loan Debt: Who is Paying it Down? Introduction 7of how it influences families’ financial These helpers are typically older and student loan repayment that we
outcomes.4 Our data allow us to have higher incomes. Third, low- observe, we explore a few possibilities
expand the existing literature in income and older borrowers are more for how targeted debt assistance
several key ways: first, we describe likely to be behind on payments or in programs could be expanded to
the burden it places on the collective deferral, and 7.1 percent of borrowers alleviate the burden of existing
household income statement. If in a given year saw an increase in student loan borrowers. As a general
parents and spouses pay a significant their balance, putting them on track principle, because the majority of
portion of monthly payments, then to never repay their loans. Finally, we borrowers are managing their debt
the economic impacts (on saving, observe large racial gaps in student without being excessively burdened,
spending, income, etc.) of student loan repayment. Compared to White efforts to alleviate undue burdens
debt may be broader than previously and Hispanic student loan borrowers, from student loan debt should be
thought and have strong policy Black borrowers are less likely to be targeted at those who are dealing with
implications (Lochner et al. 2018). making progress on their loans. genuinely challenging circumstances
Our analysis also speaks to the existing In summary, this report finds that in repayment. This is true for payment
literature focused on student loan student debt holders are not a assistance efforts like income-driven
repayment (as opposed to origination), monolithic group. Many borrowers repayment (IDR) programs as well
and the impact of debt on finances are not unreasonably burdened as more aggressive actions like debt
post-education (Goodman et al. 2019; by student loan payments and are forgiveness. A relatively easy first
Bleemer et al. 2017). We complement making payments on time. But step in expanding targeted assistance
existing work on loan progress and certain segments of the student loan would be to help additional borrowers
payoff (Gibbs 2017; Conkling and population are more substantially benefit from improved access to
Tremper 2018) by adding cuts on age, burdened by their debt, especially existing payment assistance programs,
income, and race. We provide a unique low-income borrowers, the elderly, such as IDR. Student loan debt policies
perspective on monthly payment and Black borrowers. Moreover, we and assistance programs should also
burdens since we can account for find that a significant portion of take into consideration the extent
unofficial “help” received and take- student debt payments are made not to which students rely on a network
home income, as opposed to official by the loan holder, but by parents and of people to repay their student
measures of scheduled payments spouses, who do not directly reap the loans. Loan origination programs
and taxable income (Looney and labor market returns to higher human might want to rebalance eligibility of
Yannelis 2015). Payment-to-income capital investment. This means that loans between students and parents.
ratios are an important input into the economic impacts of student debt Additionally, there could be more
income-driven repayment formulas likely affect a broader portion of the avenues for payment assistance for
(Herbst 2019), and an accessible population than previously thought. parents. A possible complement to
means of understanding how burdened repayment relief programs is to allow
What should be done to address the for restructuring or forgiveness of
the student debtor population is. disparate patterns we find in student student debt through a bankruptcy-like
Our findings are as follows. First, loan repayment? It goes without process. A fur ther step to address
although the median student loan saying that curbing the rise in tuition undue payment burdens would be to
borrower manages to make their costs and student loan debt borne expand efforts to provide targeted
scheduled payment of $2,071 or by students and their families would debt forgiveness to those most
3.8 percent of their take-home income address the problem at its root. In burdened. Targeted student loan
annually, payment burdens vary widely, addition, reducing racial gaps in debt forgiveness could be a means
with low-income and younger student income and wealth would boost of rebalancing our investments in
loan borrowers most burdened by families’ ability to pay for tuition public goods such as education across
student loan payments. Second, almost and repay student loan debt among communities and insuring against the
40 percent of individuals involved in segments of the population most risk that those investments fail to pay
student loan repayment are helping burdened by student loan debt. off for certain communities, Black and
someone else pay off their student Setting aside these structural issues Hispanic borrowers disproportionately.
loan debt, with most helpers holding that contribute to the patterns of
no student loan debt themselves.
8 Introduction Student Loan Debt: Who is Paying it Down?About the Data
Our sample is drawn from a individuals, we observe their age Chase–Experian sample oversampled
de-identified universe of 1.8 million and take-home income (based on Chase customers who made at least
families for whom we observe Experian their checking account inflows). one student loan payment between
credit bureau data for December 2015 Our main sample differs from the October 2012 and December 2013.
through November 2016, as well as nation in some important aspects. As a result, it likely oversampled
administrative checking account data, First, by construction, our sample student loan borrowers, who tend
which allow us to observe income, excludes individuals who are unbanked, to be younger (median age of 39)
student loan payments, and key roughly 6.5 percent of the nation than the typical credit bureau record
attributes of the student loan tradeline (FDIC 2018), and those who do not holder (median age of 51). However,
and the account holder. We focus on a have a credit bureau record, roughly among student loan borrowers, it likely
subset of these customers who have at 11 percent of the adult population oversampled older and actively paying
least five checking account transactions (Brevoort, Grimm, and Kambara 2015). borrowers because many borrowers
and $12,000 in deposit inflows during These populations likely overlap. are not required to make payments
our twelve-month window in order Second, our sample differs slightly until six months after leaving school.
to select customers for whom we are from national benchmarks of credit With a median student loan balance of
confident their Chase account is their bureau record holders and student loan $14,452, the Chase–Experian sample of
primary checking account. Specifically, borrowers. As shown in the Appendix, student loan borrowers has slightly less
we observe roughly 220,000 primary the age distribution of credit bureau student loan debt compared to national
account holders who have a student holders in the Chase–Experian sample benchmarks. Although the median
loan tradeline and an additional tilts in favor of younger individuals student loan borrower has two trade
80,000 primary account holders who than national benchmarks. On the lines, 16 percent of student loan holders
do not have a student loan tradeline contrary, our sample of student loan have five or more trade lines (see
but who we observe making student borrowers is slightly older than national Table A1 in Appendix). These borrowers
loan payments. For each of these benchmarks. This is likely because our also tend to have higher balances.
Table 1: The sample draws from a universe of 1.8 million families for whom we observe Chase checking account and
Experian credit bureau records
Sample of student loan holders Full Chase-Experian sample
Number of people 220,710 1,843,857
Median number of student loans 2 0
Median age 39 51
Median income $56,083 $47,541
Share female 45.5% 40.1%
Share with mortgage 61.6% 67.2%
Median liquid assets $3,621 $3,930
Median student loan balance $14,452 $0
Median installment loan balance $107,357 $194,937
Share of sample with deferred student loans 8.5% 1.4%
Share with reported student loan payment of zero 3.1% 0
Source: JPMorgan Chase Institute
Student Loan Debt: Who is Paying it Down? Introduction 9For the purposes of our analyses, we payments made via paper check sample: December 2015 through
leverage the Chase–Experian sample or money order are unable to be November 2016. We explore
to observe three different payment categorized as such. Additionally, payment burden in Finding 1.
metrics: we are unable to link electronic • Payment help given and received
• Scheduled payments are based on student loan payments out of the (Reported payment minus
Experian data and reflect the min- checking account to individual Observed payment): The relation-
imum monthly required payments trade lines. Thus, these payments ship between reported payments
for the student loan tradeline to stay are not necessarily being made and observed payments can indicate
current. These payments represent toward the borrower’s own student help given and received (Figure 1).
the promised and expected obliga- loan debt balance; that is, the When observed payments exceed
tion of the borrower as reported borrower may be making payments reported payments, we call these
by the loan servicers to Experian. towards someone else’s loan. account holders “net helpers.” In
• Reported payments are based We are able to combine these payment addition, we can observe “pure
on Experian data and indicate the metrics with additional bank and helpers” who have no student loan
total payments received by the loan Experian data to examine five key tradeline (and thus mechanically no
servicer and reported to Experian. student loan borrowing outcomes. reported payment) but nonetheless
These payments represent the total • Payment burden (payment are making student loan payments.
amount re-paid by the borrower divided by income): We define In Finding 2, we calculate the share
or anyone else during the month. payment burden as the payment of individuals involved in student
amount divided by take-home loan repayment who are extend-
• Observed payments are based ing help to another person and
on Chase checking account data income; e.g., scheduled burden is
the borrower’s scheduled payment quantify the amount of help given
and reflect the total student loan and received by age and income.
payments from the borrower’s divided by their income. To account
for seasonal fluctuations in income Specifically, we characterize people
Chase checking account toward any as having given (received) help
student loan trade line. Notably, we such as tax rebates and year-end
bonuses, we consider the sum when the reported payment is at
are only able to categorize student least one month’s payment more
loan payments as such if they are of payments and income during
the last twelve months of our (less) than the observed payment.6
made electronically. Student loan
Figure 1: Individuals involved in student loan repayment can be paying down their own debt or helping someone else repay
their student loan debt
Pure helpers
Individuals who do not have a student loan tradeline in their These Individuals are helping
na e but have ade pay ents towards student loans someone else pay down their
student loan debt in that they are
aking student loan pay ents out
Loan holders – net helpers of their checking account but do not
have student loan debt or have
Individuals who have a student loan tradeline in their na e;
lower reported student loan
their student loan pay ents out of their checking account pay ents according to Experian
exceeds what the credit bureau reports. This excess is helping credit bureau data.
pay down another person’s student loan
Loan holders – payers
Individuals in these groups Individuals who have a student loan tradeline in their na e and
hold student loan debt have ade student loan pay ents out of their checkings
account but are net helpers. A subset of these individuals might
be receiving help fro others to
the extent that their reported
Loan holders – non-payers
pay ents exceed their
Individuals who have a student loan tradeline in their na e observed pay ents.
but have not ade pay ents towards student loans out of
their checking account.
Source: JPMorgan Chase Institute
10 Introduction Student Loan Debt: Who is Paying it Down?When reported payments exceed values denote pre-payment by same rate going forward. Finally,
observed payments, this disparity the borrower or payments in we calculate how many years it will
can have two interpretations. First, it excess of the minimum required. take for the borrower to zero out
can indicate that another method of • Deferment or forbearance: the balance on all their student
payment is being used to pay down the Deferment and forbearance can loans, holding income constant.
loan besides electronic withdrawals allow borrowers to temporarily In Finding 3 we measure progress
from the borrower’s Chase checking stop making payments on their on student loan repayment in three
account. The borrower is making loan for up to three years under a ways. First, we look at each borrower’s
student loan payments via paper variety of circumstances, including payment shortfall, or how much of
checks or money orders, which we are returning to school, medical their scheduled payments they did
unable to categorize as such, or via circumstances, and economic not end up fulfilling. Second, we
payments out of a non-Chase account. hardships.8 During this time, loan calculate how many borrowers are
The second possibility is that the balances may increase if borrowers in deferral. And third, we calculate
borrower is receiving help. Someone do not make interest payments. how long a borrower will take to
besides the borrower—such as a Here we include any borrower in completely pay off their student debt
parent, a spouse, or a child—is making the “deferred” group if any of their given current payment levels.9
payments directly to the loan servicer tradelines are flagged by Experian
on the borrower’s behalf. We explore In Finding 4, we compare these student
as being deferred or in forbear- loan borrowing outcomes among
help received and given by borrowers ance. For simplicity, and because
in Finding 2 by calculating the differ- Black, Hispanic, and White account
less than 1 percent of these holders for a subsample of roughly
ence between reported and observed people are in forbearance alone
payments for each borrower. It is worth 110,000 primary account holders
without deferral, we will exclusively in the Chase–Experian sample for
noting that since we do not observe refer to them as in “deferral.”
the identity of who makes payments on whom we observe self-reported race.
any given tradeline, we cannot make • Projected time to payoff: We Of these, 12,154 have a student loan
conclusions, except in the aggregate, project the time to payoff for both tradeline. We take this self-reported
about who is providing help to whom. borrowers in active repayment race information from the data asset
and those in deferral. To do this, described in Farrell et al. (2020),
• Payment shortfall and prepay- we impute the interest rate for where we observe self-reported race
ment: We calculate payment each tradeline in our data using for Chase account holders obtained
shortfall as scheduled payments month-to-month changes in through voter registration records in
minus reported payments, divided balances together with reported Florida, Georgia, and Louisiana from
by the average of scheduled pay- payments. For example, if a 2018. Because voter registration forms
ments for the year.7 The numerator tradeline’s balance in January was do not separately ask about race and
is simply the difference in dollars $1,000, the borrower’s reported ethnicity, we are unable to separately
between what a borrower should payment was $100, and the balance analyze race and Hispanicity (e.g. we
have paid and what they actually in February was $910, we assume cannot distinguish Hispanic individuals
paid. By dividing this difference the interest charge for January who identify as White from Hispanic
by average scheduled payment, was $10. Next, we average the individuals who identify as Black). For
we change the unit of measure interest rate across the borrower’s this reason, we use the word “race” as
from dollars to months’ worth of tradelines, weighting by the current a shorthand to describe responses to
payments. Positive values repre- balance. We then take the average the question on the voter registration
sent how many months behind reported payments made in our form, acknowledging that many people
the borrower is, and negative sample year and suppose the consider Hispanic identity an ethnic
borrower makes payments at the category and not a racial group.
Student Loan Debt: Who is Paying it Down? Introduction 11Finding One
Although the median student loan of $3,936. The fact that, in aggregate, (borrower’s annual payment divided by
borrower is obligated to pay 3.8 per- reported payments skew slightly higher borrower’s annual income). Scheduled
cent of their take-home income, than scheduled payments suggests that and reported burden are similar,
many borrowers, especially lower- the typical borrower is making pay- as in the left panel of Figure 2, with
income and younger borrowers, ments on track or ahead of schedule. median payment burdens of 3.8 and
face burdens well over 10 percent. We further examine progress on loans 3.9 percent of take-home income,
The left panel of Figure 2 shows the on a per person basis in Finding 3. respectively. The 75th percentile of
10th, 25th, 50th, 75th and 90th per- Observed payments skew lower than scheduled burden is 7.3 percent, imply-
centiles of scheduled, reported, and both scheduled and reported payments, ing that a quarter of borrowers are
observed payment amounts for all with a median of only $1,594. This gap obligated to pay at least 7.3 percent of
student debt holders in our sample could signify that borrowers are either their take-home pay, while 10 percent
over the twelve-month period. The making additional payments that we do of the sample is obligated to pay
distribution of median scheduled and not observe via paper checks or non- at least 13.3 percent (the 90th per-
reported payments are similar, with a Chase accounts or receiving payment centile) of their take-home pay.
median of about $2,070 and a 75th per- help on their loans from others. Similar to the left panel of Figure 2,
centile of $3,684. Reported payments observed burden is markedly lower
tend to skew slightly higher, with a We explore this further in Finding 2. than scheduled or reported burden,
median of $2,146 and a 75th percentile The right panel of Figure 2 shows with a median of only 2.7 percent.
the distributions of payment burdens
Figure 2: Distributions of annual payment level and burden by payment type
Distribution of annual payment amounts Distribution of payment burdens
(Payment as a percentage of income)
90th
Percentile 90th
$7,000 14% Percentile
$6,000 12%
$5,000 10%
$4,000 75th 8%
Percentile 75th
Percentile
$3,000 6%
50th
$2,000 Percentile 4% 50th
Percentile
25th
Percentile 2% 25th
$1,000 Percentile
10th Percentile
10th Percentile
$0 0%
Scheduled Reported Observed Scheduled Reported Observed
payment payment payment burden burden burden
Note: Scheduled payment is the sum of required minimum monthly payments for the twelve-month sample period November 2015 through December 2016. Reported
payment is the sum of all payments made against the borrower's student loans during the sample period. Observed payment is the sum of all payments made out of the
borrower's Chase accounts during the sample period. Income refers to take-home income.
Source: JPMorgan Chase Institute
12 Finding One Student Loan Debt: Who is Paying it Down?Next we examine how annual payment hand, as shown by Looney and Yannelis similar when we focus on scheduled or
levels and burden vary by age and (2015), higher income borrowers tend observed payments. Most notably, the
income. These borrower attributes to have higher debt balances. On the median payment amount is relatively
are of great interest and often not other hand, higher income borrowers constant across income groups. The
present in other administrative may be better able to pay off their median borrower making $30,000
datasets. Age speaks to the extent to student loan debt completely, and pays $1,605 toward student loans,
which borrowers are experiencing the income-driven repayment programs while the median borrower making
strain of student loan debt repayment aim to reduce the payment burden $117,000 pays $2,700, a difference
over the life cycle and potentially on among low-income borrowers. of only $91.25 per month. However,
behalf of others rather than their own Figure 3 plots the 25th, 50th, 75th, and extremely large payments, captured
education. Income allows us to see 90th percentiles of reported payments in the 90th percentiles, are far more
whether debt loads are higher among by income and age. We focus on likely for high-income borrowers,
people with higher earning power. reported payments in order to capture with those making at least $30,000
It is ambiguous whether we might payments made, but the relationship per year paying $5,038 and those
expect to see higher or lower levels of between payments and income looks making $130,000 paying $9,760.
payment levels by income. On the one
Figure 3: Payment levels by income and age
Distribution of annual reported pa ment b income Distribution of annual reported pa ment b age
$15,000 $15,000
90th Percentile
$10,000 $10,000
75th Percentile 90th Percentile
$5,000 $5,000
50th Percentile 75th Percentile
25th Percentile 50th Percentile
25th Percentile
$0 $0
20 40 60 80 100 200 30 40 50 60 70
Income ($1,000s) Age
No e: Percen iles are calcula ed wi hin wen y income and age quan iles, respec ively. Each bin is represen ed along he x-axis by i s average value. Repor ed paymen is
he sum of all paymen s made agains he borrower's s uden loans during he welve-mon h sample period November 2015 hrough December 2016. Income refers o
ake-home income.
Source: JPMorgan Chase Ins i u e
Student Loan Debt: Who is Paying it Down? Finding One 13In Figure 4, we explore how payment reported burden increases. Again, this The median scheduled burden for
burdens vary by age and income.10 gap could signify that older borrowers the lowest income group (around
Median burdens are highest among pay primarily with paper checks or $16,000 annual take-home income)
people in their 20s (that is, when non-Chase accounts, or it could signify is 11.5 percent and for the highest
most people have just graduated that order borrowers receive a large income group ($250,000 annual
from college and enter the labor amount of help on their loans. We take-home income) is 1.5 percent.
force) and lowest for people in their explore this fur ther in Finding 2. Observed burdens are also negatively
late 30s. Starting with 40-year-olds, The left panel of Figure 4 shows that correlated with income, but observed
the median scheduled and reported scheduled, reported, and observed burden is significantly lower than
burdens steadily increase with age. payment burdens are strongly other burden metrics for low-income
Observed burden is again system- negatively correlated with income. borrowers. Again, this could be due to
atically lower than scheduled and This is not surprising in light of the fact unobserved payments or help received
reported burdens. Most noteworthy, that Figure 3 shows little variation in from others. Next we turn to examine
however, is that observed burden reported payment values by income. the extent of help being received and
steadily decreases with age, while given in student loan repayment.
Figure 4: Median payment burden by income and age
Median payment burdens by income Median payment burdens by age
(Payment as a percentage of income) (Payment as a percentage of income)
12% 12%
10% 10%
8% 8%
6% 6%
4% 4%
2% 2%
0% 0%
20 40 60 80 100 200 30 40 50 60 70
Income ($1,000s) Age
Reported Scheduled Observed
Note: Percentiles are calculated within twenty income and age quantiles, respectively. Each bin is represented along the x-axis by its average value. Scheduled payment is
the sum of required minimum monthly payments for the twelve-month sample period November 2015 through December 2016. Reported payment is the sum of all
payments made against the borrower's student loans during the sample period. Observed payment is the sum of all payments made out of the borrower's Chase
accounts during the sample period. ncome refers to take-home income.
Source: JPMorgan Chase nstitute
14 Finding One Student Loan Debt: Who is Paying it Down?Finding Two
Almost 40 percent of individuals own student loan debt (220,710 people slightly larger than the observed
involved in student loan repay- or 12 percent of the Chase–Experian payments made by student loan holders
ment are helping someone else sample) or because they are making themselves ($1,594).11 Notably, student
pay off their student loan debt, payments on someone else’s student loan borrowers have higher median
with most helpers holding no loan debt (80,873 people or 4 percent reported payments ($2,146) than
student loan debt themselves. of the Chase–Experian sample). It shows observed payments ($1,594). In fact,
Figures 2 and 3 above show that, in that 39 percent of individuals involved 42 percent of student loan borrowers
aggregate, the distribution of observed in student loan repayment are actually have higher reported payments than
payments skews lower than the distribu- helping someone else repay their loans observed payments, additional evidence
tion of reported payments. This suggests as either “pure helpers” (27 percent) that many borrowers are receiving help
that more money is being received by or “net helpers” (12 percent). Most of from others with their obligations.
student loan servicers than we observed those helpers (69 percent) are pure Table 2 further breaks down the sample
leaving borrowers’ checking accounts. helpers who do not have a student of borrowers according to whether
This additional money could be coming loan tradeline in their own name. The they are making payments or not.
from relatives or friends of the borrower remaining 31 percent of helpers are net Fifty-nine percent of our borrowers
in the form of financial help. We explore helpers—they have a tradeline in their are “Payers” in that they are making
this possibility in two ways. First, we name but their observed payments payments less than or roughly equal to
look for these potential “helpers” in exceed payments reported to Experian, their reported payments. Even among
our data; specifically, we look for “pure suggesting they are helping to make pay- this group, 34 percent show larger
helpers”—people who make payments to ments on another person’s tradelines. reported payments than observed
student loan servicers but do not have Compared to the median student loan payments, potentially indicating some
a student loan according to Experian holder, helpers of both sorts tend to level of payment help received. Finally,
data—and “net helpers”—people who are have higher incomes (over $70,000 the remaining 25 percent of student
making larger student loan payments among helpers compared to $54,000 loan borrowers are “non-payers” in that
than what is received on their own among all borrowers). Pure helpers tend we observed no student loan payments
tradeline. In order to be classified as a to be substantially older than the typical were made from their Chase accounts
net helper, a borrower must make at student loan holder (47 compared to 39, during our twelve-month observation
least one month’s worth of their own respectively), suggesting that some of window. Non-payers tend to be older
payments to someone else’s tradeline. them could be parents making payments and have lower incomes than the rest
Second, we examine the magnitude on behalf of their adult children. Net of student loan holders. However,
of help by calculating the difference helpers tend to be slightly younger 88 percent of these borrowers still had
between reported and observed (37 years) than the median student payments made against their tradelines
payments for each individual in our data loan holder, suggesting that some of by someone. Moreover, the median
to explore any systematic trends in who these helpers could be spouses. reported payment for the non-paying
is receiving and giving financial help. group is $1,783, only $363 lower than
Table 2 lists summary statistics for The degree of this help is substantial.
The median size of observed payments the median student loan payment for all
everyone in our data involved in student student loan holders, suggesting that the
loan repayment either through their by pure helpers ($1,772) is actually
amount of outside help is substantial.
Student Loan Debt: Who is Paying it Down? Finding Two 15Table 2: Summary statistics for accounts associated with student debt
No student loan
tradeline but
makes payments Has student loan tradeline
Non-paying
Pure helpers Net helpers Paying debtors All
debtors
e.g., Debtors, e.g., Paying e.g. Debtors in Anyone with
Parents or family
who are also Students, deferment, grace an open
members making
Description helping make paying parents period, IDR with student loan
student loan
student loan on parent-plus $0 payment, or tradeline
payments on
payments on loan, excluding delinquent according to
student’s behalf
another’s behalf net helpers Experian data
Number 80,873 36,495 129,946 54,269 220,710
Share of full Chase-
4% 2% 7% 3% 12%
Experian sample
Share of individuals
involved in student 27% 12% 43% 18% 73%
loan repayment
Share of student
N/A 17% 59% 25% 100%
loan holders
Median age 46.9 36.9 37.9 44.9 38.9
Median annual income 75,568 72,919 53,818 44,613 54,083
Median student
$0 $10,812 $15,522 $14,370 $14,451
debt balance
Median observed
$1,772 $4,184 $2,052 $0 $1,594
payments (annual)
Median reported
$0 $1,854 $2,375 $1,783 $2,146
payments (annual)
Share of people with
reported student loan
0% 0% 34% 88% 42%
payments > observed
student loan payments
Share of people with
no reported student N/A 1% 0% 12% 3%
loan payments
Source: JPMorgan Chase Institute
16 Finding Two Student Loan Debt: Who is Paying it Down?In Figure 5, we show the distribution repayment help from another person. likely to receive larger amounts of help
of different repayment roles by age We explore this possibility below. and are unlikely to extend help. The
and income band. More than a third The nature of our data do not allow median and 25th percentile of help is
of the lowest-income individuals us to determine whose tradelines the zero or near zero for all income groups
are not making payments on their helpers are contributing to, but we below $81,000, with the exception
student loans. In contrast, more can describe the demographics of of the lowest group, whose median
than half of the highest-income people who appear to be receiving or person receives approximately $200
individuals are pure helpers (40%) giving help, which we do in Figure 6 by of help. Of borrowers with around
or net helpers (17%) towards comparing person-level calculations $30,000 of income, 25 percent
loans held by other individuals. of reported payments minus observed receive at least $1,000 of help.
By age, we observe among individuals payments among student loan
under 35 years old the modal person borrowers. A positive value indicates
is actively repaying their loans. In the borrower could be receiving
contrast, among the oldest age groups, help, while negative values indicate Older individuals
roles diverge. Those in the 65 plus borrowers who must be providing are more likely than
category are most likely to be either help in addition to any payments younger borrowers to
helping with someone else’s loan they make on their own loans. both extend and receive
repayment (35 percent are pure help- The left panel of Figure 6 plots help help with student
ers) or not making payments on their given and received across the income loan repayment.
own loan (33 percent are non-paying distribution. Perhaps unsurprisingly,
debtors). These non-paying debtors lower-income borrowers are more
could be simply not paying or receiving
Figure 5: Share of non-debt holding helpers and debt-holding payers by income and age group
Distribution of payer types by income quintile Distribution of payer types by age group
10.8%
17.0% 1 .9% 18.2%
24.9% 25.5%
33.5% 13.6% 38.3% 34.9%
37.1%
7.5% 39. %
9.7% 15.7%
11.5% 12.2%
6.5%
14.4% 8.8% 9.4%
45.2%
17.4%
57.8% 24.1%
51.1%
47.2% 51.5% 47.1% 29.6%
34.4%
39.4%
32.5%
34.5%
30.3%
20.2% 23.9%
16.4% 17.9% 15.3% 18.5%
12. % 10.3% 14.6%
Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 18-24 25-34 35-44 45-54 55-64 65+
(Figure 6: Distribution of help given (negative values) and received (positive values) by income and age
Distribution of payment help received by borrower income Distribution of payment help received by borrower age
$2,500 $2,500 75th
Percentile
$2,000 $2,000
Net payment help received
Net payment help received
$1,500 $1,500
$1,000 75th $1,000
Help received Help received 50th
Percentile Percentile
$500 $500
50th 25th
$0 Percentile $0
Percentile
Help given Help given
-$500 -$500
-$1,000 25th -$1,000
Percentile
-$1,500 -$1,500
20 40 60 80 100 200 30 40 50 60 70
Income ($1,000s) Age
Note: Percentiles are calculated within twenty income and age quantiles, respectively. Each bin is represented along the x-axis by its average value. Payment help is the
difference between reported and observed payments; that is, the difference between the total paid against a borrower's loans and the total payments made toward student
debt from the borrower's Chase accounts. Negative payment help represents help provided by the borrower to other student debt holders. Income refers to take-home
income.
Source: JPMorgan Chase Institute
At higher income levels, few bor- borrowers could be from their child The results, shown in Appendix
rowers receive any help, and a who is helping to pay down Parent Figure A1 are qualitatively similar to
significant portion of high-income PLUS loans that were taken out on for Figure 6, with help received increasing
borrowers are net givers of help the benefit of the child’s education. markedly for borrowers above age 50.
despite holding loans themselves. This scenario is corroborated by the In summary, we find that a large
The medians and 75th percentiles for fact that, based on the age of the group of people are assisting student
income above $134,000 are all zero, borrower at loan origination, most loan borrowers with repayment,
and 25 percent of borrowers paid loans held by people over 60 appear the majority of whom do not have a
at least $594 toward student loans to be loans taken out for children’s student loan tradeline in their own
over and above what was credited education rather than their own edu- name. This underscores the extent
toward their own debt. This additional cation: 75 percent of 60+ borrowers to which student loan repayment is a
help could reflect a high-income originated their first student loan after “family affair,” perhaps more so than
borrower paying down student loan the age of 40, and the modal 60+ previously thought. Moreover, the help
debt for a spouse, parent (e.g. a borrower originated their first loan at appears to involve intergenerational
Parent PLUS loan), or child. However, age 55.12 Additionally, we observed a transfers in both directions: we
this last possibility is less likely in large number of non-payers among observe younger individuals serving
light of the right panel of Figure 6. student loan borrowers in the 55–64 as net helpers by servicing their own
The right panel of Figure 6 plots the and 65+ age bins in Figure 5. debt and the debt of others, many
distribution of help by age. It shows One caveat is that it is also possible older borrowers who are making no
that older individuals are more likely that help received particularly among payments on loans in their own name
to receive help. The 25th percentiles older individuals could reflect unob- but are receiving help from others,
and medians are all approximately served payments made out of a non- and many older individuals with no
zero for borrowers below age 60. Chase account or via paper checks.13 loan in their own name nonetheless
The 75th percentiles hover around To check for this, we re-calculate the making student loan payments.
$500 from the youngest bins through right panel of Figure 6 restricting While helpers tend to have higher
the early 40s, after which the values to people who made at least one incomes, borrowers receiving help
increase to a maximum of $2,348 electronic student loan payment are likely to be lower-income and
for the oldest borrowers. This between 2013 and 2016. Individuals older. Thus, older individuals are
matches the increase in median help who have previously setup electronic more likely than younger borrowers
received for borrowers over 60. The payments are less likely to use paper to both extend and receive help
large amount of help going to older checks than the general population. with student loan repayment.
18 Finding Two Student Loan Debt: Who is Paying it Down?Finding Three
Low-income and older borrowers are The left panel of Figure 7 shows The right panel of Figure 7 shows pay-
more likely to be behind on payments payment shortfall in our sample year ment shortfall by age bin. Again, most
or in deferral, and roughly 7 percent by income bin. The vast majority of groups have little shortfall, as all medians
of borrowers are projected not to borrowers are not behind on their pay- and 75th percentiles are at or near zero.
repay their loans. Having documented ments, but low-income borrowers are However, grouping by age instead of
the large degree of financial help more likely to be behind on their pay- income reveals that we see many people
given and received in student loan ments. Across the income spectrum, who are behind in payments, as well as
repayment, we next turn to examining the 25th, 50th, and 75th percentiles few who are making significant pre-
the extent to which individuals are of borrowers are paying on schedule payments. Twenty-five percent of
making progress on paying down their or are less than one month behind. borrowers under 30 years old have a
debt. We measure progress on student However, 10 percent of borrowers (the pre-payment (a negative shortfall) of
loan repayment in three ways. First, 90th percentile) with incomes less at least one month. At the same time,
we look at each borrower’s payment than $30,000 in take-home income 10 percent of borrowers under 30 are
shortfall, or how much their reported are 4 to 6 months or more behind on at least two months behind. Shortfalls
payments fall short of their scheduled their payments in just one year, and at the 90th percentile stay around
payments within our sample year. 10 percent of middle-income borrow- two months up to age 40, while the
Because reported payments include ers (between $30,000 and $50,000) 25th percentiles move toward zero.
help received from others, this measure also have shortfalls of at least two This could be due to selective survival:
of shortfall takes into consideration all months. These results largely align those who pre-pay their loans in their
outside help. Second, we calculate how with conventional wisdom that low- 20s are done paying by the end of
many borrowers are (temporarily) in er-income borrowers are more likely to their 30s and thus disappear from our
deferral or forbearance. And third, we have trouble paying. This is especially sample of student loan holders. After
calculate how long a borrower will take true considering Figure 3, which shows age 45, shortfalls increase markedly,
to completely pay off their student debt that scheduled payments are largely with the 90th percentiles rising above
given their current payment levels. constant across income groups. three months and the 75th percentiles
rising as well to about 0.25 months.
Figure 7: Payment shortfall by income and age
Distribution of payment shortfall by borrower income Distribution of payment shortfall by borrower age
7 7
Payment shortfall in months
Payment shortfall in months
6 6
5 5
4 4
3 3 90th Percentile
2 2
1 90th Percentile 1 75th Percentile
75th Percentile 50th Percentile
0 0
50th Percentile 25th Percentile
−1 25th Percentile −1
−2 −2
20 40 60 80 100 200 30 40 50 60 70
Income ($1,000s) Age
Note: Percentiles are calculated within twenty income and age quantiles, respectively. Each bin is represented along the x-axis by its average value. Payment shortfall is
the difference between all scheduled and reported payments during the tweleve-month sample period December 2015 through November 2016, divided by average
monthly scheduled payment. Negative values of shortfall constitute pre-payment. Income refers to take-home income.
Source: JPMorgan Chase Institute
Student Loan Debt: Who is Paying it Down? Finding Three 19You can also read
