the effects of student loans on long-term household...

31
The Effects of Student Loans on Long-Term Household Financial Stability Dora Gicheva 1 Jeffrey Thompson 2 1 UNC Greensboro 2 Federal Reserve Board October 26, 2013 Gicheva and Thompson Student Loans and Financial Stability October 26, 2013 1 / 26

Upload: others

Post on 14-Jun-2020

1 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: The Effects of Student Loans on Long-Term Household ...edpolicy.umich.edu/...thompson-presentation-effects-student-loans.pdf · The E ects of Student Loans on Long-Term Household

The Effects of Student Loans on Long-Term HouseholdFinancial Stability

Dora Gicheva1 Jeffrey Thompson2

1UNC Greensboro

2Federal Reserve Board

October 26, 2013

Gicheva and Thompson Student Loans and Financial Stability October 26, 2013 1 / 26

Page 2: The Effects of Student Loans on Long-Term Household ...edpolicy.umich.edu/...thompson-presentation-effects-student-loans.pdf · The E ects of Student Loans on Long-Term Household

Introduction

Motivation

Unintended consequences of student loan program that we should beaware of

Need to take into account all costs for cost-benefit analysis

Less is known about long-term consequences of holding student debt

Gicheva and Thompson Student Loans and Financial Stability October 26, 2013 2 / 26

Page 3: The Effects of Student Loans on Long-Term Household ...edpolicy.umich.edu/...thompson-presentation-effects-student-loans.pdf · The E ects of Student Loans on Long-Term Household

Introduction

Motivation

Are growing debt burdens in the right tail of the distributionmanageable?

I Should not ignore positive return to student borrowing

What about borrowers who do not complete degree?

Gicheva and Thompson Student Loans and Financial Stability October 26, 2013 3 / 26

Page 4: The Effects of Student Loans on Long-Term Household ...edpolicy.umich.edu/...thompson-presentation-effects-student-loans.pdf · The E ects of Student Loans on Long-Term Household

Introduction

Research Questions

How do student borrowers fare financially after graduation, comparedto similar non-borrowers?

Are borrowers who do not complete Bachelor’s degree more likely toexperience financial hardship?

How can we address bias from omitted variables andsimultaneity/reverse causality?

Gicheva and Thompson Student Loans and Financial Stability October 26, 2013 4 / 26

Page 5: The Effects of Student Loans on Long-Term Household ...edpolicy.umich.edu/...thompson-presentation-effects-student-loans.pdf · The E ects of Student Loans on Long-Term Household

Introduction

Empirical Approach

Data from Survey of Consumer Finances

Student loans that originated between 1970 and 2010

Observe outcomes in 1995-2010

Instrument for amount borrowed using variations in aggregateborrowing trends

I Exogenous policy-induced changes in eligibility and take-up

Gicheva and Thompson Student Loans and Financial Stability October 26, 2013 5 / 26

Page 6: The Effects of Student Loans on Long-Term Household ...edpolicy.umich.edu/...thompson-presentation-effects-student-loans.pdf · The E ects of Student Loans on Long-Term Household

Introduction

Overview of Findings

Conditional on broad measures of educational attainment, studentdebt linked to credit constraints and bankruptcy

I $10,000 in student loans increases probability of being turned down bycreditors by about 6 percentage points

Home ownership rates may also be affected

Relationship stronger for non-completers

Gicheva and Thompson Student Loans and Financial Stability October 26, 2013 6 / 26

Page 7: The Effects of Student Loans on Long-Term Household ...edpolicy.umich.edu/...thompson-presentation-effects-student-loans.pdf · The E ects of Student Loans on Long-Term Household

Introduction

Post-Graduation Outcomes Affected by Student Debt:Existing Evidence

Graduate school attendance (mixed evidence)

Career choice and starting salary

Specialized studies:I Physician specialty choice (limited impact)I Law school graduates’ choice to enter public sector law

Entry into marriage

Gicheva and Thompson Student Loans and Financial Stability October 26, 2013 7 / 26

Page 8: The Effects of Student Loans on Long-Term Household ...edpolicy.umich.edu/...thompson-presentation-effects-student-loans.pdf · The E ects of Student Loans on Long-Term Household

Methodology Specification

Empirical Model

Want to estimateE (Y ) = α(Loans) + Xβ

Omitted variable bias:

I Student loans correlated with many unobservables that contribute tofinancial stability

I Type of education, academic success, family background,...

Possible reverse causality

Gicheva and Thompson Student Loans and Financial Stability October 26, 2013 8 / 26

Page 9: The Effects of Student Loans on Long-Term Household ...edpolicy.umich.edu/...thompson-presentation-effects-student-loans.pdf · The E ects of Student Loans on Long-Term Household

Methodology Specification

Empirical Model

Two-stage least squares approach

First stage:E (Loans) = Xδ + γZ

The instrument Z equals average amount borrowed per full-timeequivalent student (including non-borrowers) in constant dollars inyear when respondent was 17 years old

I From Trends in Student Aid 2012

Gicheva and Thompson Student Loans and Financial Stability October 26, 2013 9 / 26

Page 10: The Effects of Student Loans on Long-Term Household ...edpolicy.umich.edu/...thompson-presentation-effects-student-loans.pdf · The E ects of Student Loans on Long-Term Household

Methodology Specification

Aggregate Borrowing Trends

Reauthorizations of the HEA of 1965

I New types of aid

I Expand loan limits and eligibility

Interest rate changes

Private loan market

Increasing gap between costs and grant aid

Information availability

Gicheva and Thompson Student Loans and Financial Stability October 26, 2013 10 / 26

Page 11: The Effects of Student Loans on Long-Term Household ...edpolicy.umich.edu/...thompson-presentation-effects-student-loans.pdf · The E ects of Student Loans on Long-Term Household

Methodology Specification

Aggregate Borrowing Trends

$0

$1,000

$2,000

$3,000

$4,000

$5,000

$6,000

$7,000

$8,000

Fede

ral/Private Loan

s per FTE

 Stude

nt

Year

Federal Loans per FTE Student

Total (Federal+Private) Loansper FTE Student

Gicheva and Thompson Student Loans and Financial Stability October 26, 2013 11 / 26

Page 12: The Effects of Student Loans on Long-Term Household ...edpolicy.umich.edu/...thompson-presentation-effects-student-loans.pdf · The E ects of Student Loans on Long-Term Household

Methodology Specification

Aggregate Borrowing as Instrument

Captures exogenous changes in eligibility and take-up rates

Relationship between prevalent borrowing behavior in year of highschool graduation and individual’s borrowing decisions

Gicheva and Thompson Student Loans and Financial Stability October 26, 2013 12 / 26

Page 13: The Effects of Student Loans on Long-Term Household ...edpolicy.umich.edu/...thompson-presentation-effects-student-loans.pdf · The E ects of Student Loans on Long-Term Household

Methodology SCF Data

Survey of Consumer Finances Sample

Six cross-sectional surveys conducted between 1995 and 2010I Full student borrowing historiesI Very detailed information about assets and liabilities

Respondents born in 1954 or later - entered college after 1970

Age 29 or older

N = 12, 037

Weights make sample nationally representative

Gicheva and Thompson Student Loans and Financial Stability October 26, 2013 13 / 26

Page 14: The Effects of Student Loans on Long-Term Household ...edpolicy.umich.edu/...thompson-presentation-effects-student-loans.pdf · The E ects of Student Loans on Long-Term Household

Methodology SCF Data

Student Loans in the SCF

All debt accumulated by household members

Exact amount and year of each education loan

Adjusted for inflation

Gicheva and Thompson Student Loans and Financial Stability October 26, 2013 14 / 26

Page 15: The Effects of Student Loans on Long-Term Household ...edpolicy.umich.edu/...thompson-presentation-effects-student-loans.pdf · The E ects of Student Loans on Long-Term Household

Methodology SCF Data

Student Loans in the SCF

0.0%

5.0%

10.0%

15.0%

20.0%

25.0%

30.0%

1995 1998 2001 2004 2007 2010$0.0

$5.0

$10.0

$15.0

$20.0

$25.0

$30.0

$35.0

$40.0

Share of PEU with any atudent loan debt

Average borrowing ($1,000) among those with debt

Borrowing is in constant 2010 dollars.Gicheva and Thompson Student Loans and Financial Stability October 26, 2013 15 / 26

Page 16: The Effects of Student Loans on Long-Term Household ...edpolicy.umich.edu/...thompson-presentation-effects-student-loans.pdf · The E ects of Student Loans on Long-Term Household

Methodology SCF Data

Outcomes of Interest

Whether denied credit, granted less credit than applied for initially, ordid not apply because feared rejection - previous 5 years

At least one late payment of bills (60 days or more) - previous 5 years

Payment to income ratio over 40%I All loansI Excluding education loansI Excluding education loans and mortgage

Whether filed for bankruptcy - previous 10 years

Whether owns primary residence

Gicheva and Thompson Student Loans and Financial Stability October 26, 2013 16 / 26

Page 17: The Effects of Student Loans on Long-Term Household ...edpolicy.umich.edu/...thompson-presentation-effects-student-loans.pdf · The E ects of Student Loans on Long-Term Household

Methodology SCF Data

Summary Statistics, Outcome Variables (N = 12,037)

36.76%

8.56% 11.40% 10.77%

2.15%8.29%

60.78%

Gicheva and Thompson Student Loans and Financial Stability October 26, 2013 17 / 26

Page 18: The Effects of Student Loans on Long-Term Household ...edpolicy.umich.edu/...thompson-presentation-effects-student-loans.pdf · The E ects of Student Loans on Long-Term Household

Methodology SCF Data

Summary Statistics, Explanatory Variables (N = 12,037)

College degree 0.198Masters degree 0.061Doctorate 0.012

Female respondent 0.559White 0.683Black 0.141Hispanic 0.111Age 39.5

(7.2)

Any college-age kids in PEU (18-24) 0.109Disabled (either R or SP) 0.065

Normal income ($1,000) 86.7(211.8)

Predicted wage ($1,000) 51.6(65.2)

Gicheva and Thompson Student Loans and Financial Stability October 26, 2013 18 / 26

Page 19: The Effects of Student Loans on Long-Term Household ...edpolicy.umich.edu/...thompson-presentation-effects-student-loans.pdf · The E ects of Student Loans on Long-Term Household

Methodology SCF Data

Aggregate and Derived Controls

Annual average county-level unemployment rate from the BEA

County per-capita personal income, relative to national average

Predicted wage income

I Using occupation, demographics, full-time or part-time andself-employment status, years of education

I Current Population Survey data

Gicheva and Thompson Student Loans and Financial Stability October 26, 2013 19 / 26

Page 20: The Effects of Student Loans on Long-Term Household ...edpolicy.umich.edu/...thompson-presentation-effects-student-loans.pdf · The E ects of Student Loans on Long-Term Household

Results Main Sample

Results (Second Stage)

Denied Credit 0.0063

0.0068 0.0082 0.0073 0.0060

***

*** *** ** **

Bankrupt 0.0068

0.0071 0.0074 0.0079 0.0072

***

*** *** *** **

Late Payments 0.0036

0.0037 0.0041 0.0027 0.0023

**

** ***

Home Owner -0.0018

-0.0023 -0.0055 -0.0035 -0.0021**

Education X X X XDisabled X X XCollege-age kids X X XNormal income X XCounty UR X XCounty income X XPredicted wage X

Gicheva and Thompson Student Loans and Financial Stability October 26, 2013 20 / 26

Page 21: The Effects of Student Loans on Long-Term Household ...edpolicy.umich.edu/...thompson-presentation-effects-student-loans.pdf · The E ects of Student Loans on Long-Term Household

Results Main Sample

Results (Second Stage)

Denied Credit 0.0063 0.0068

0.0082 0.0073 0.0060

*** ***

*** ** **

Bankrupt 0.0068 0.0071

0.0074 0.0079 0.0072

*** ***

*** *** **

Late Payments 0.0036 0.0037

0.0041 0.0027 0.0023

** **

***

Home Owner -0.0018 -0.0023

-0.0055 -0.0035 -0.0021**

Education X

X X XDisabled X X XCollege-age kids X X XNormal income X XCounty UR X XCounty income X XPredicted wage X

Gicheva and Thompson Student Loans and Financial Stability October 26, 2013 20 / 26

Page 22: The Effects of Student Loans on Long-Term Household ...edpolicy.umich.edu/...thompson-presentation-effects-student-loans.pdf · The E ects of Student Loans on Long-Term Household

Results Main Sample

Results (Second Stage)

Denied Credit 0.0063 0.0068 0.0082

0.0073 0.0060

*** *** ***

** **

Bankrupt 0.0068 0.0071 0.0074

0.0079 0.0072

*** *** ***

*** **

Late Payments 0.0036 0.0037 0.0041

0.0027 0.0023

** ** ***

Home Owner -0.0018 -0.0023 -0.0055

-0.0035 -0.0021

**

Education X X

X X

Disabled X

X X

College-age kids X

X X

Normal income X

XCounty UR X XCounty income X XPredicted wage X

Gicheva and Thompson Student Loans and Financial Stability October 26, 2013 20 / 26

Page 23: The Effects of Student Loans on Long-Term Household ...edpolicy.umich.edu/...thompson-presentation-effects-student-loans.pdf · The E ects of Student Loans on Long-Term Household

Results Main Sample

Results (Second Stage)

Denied Credit 0.0063 0.0068 0.0082 0.0073

0.0060

*** *** *** **

**

Bankrupt 0.0068 0.0071 0.0074 0.0079

0.0072

*** *** *** ***

**

Late Payments 0.0036 0.0037 0.0041 0.0027

0.0023

** ** ***

Home Owner -0.0018 -0.0023 -0.0055 -0.0035

-0.0021

**

Education X X X

X

Disabled X X

X

College-age kids X X

X

Normal income X XCounty UR X

X

County income X

XPredicted wage X

Gicheva and Thompson Student Loans and Financial Stability October 26, 2013 20 / 26

Page 24: The Effects of Student Loans on Long-Term Household ...edpolicy.umich.edu/...thompson-presentation-effects-student-loans.pdf · The E ects of Student Loans on Long-Term Household

Results Main Sample

Results (Second Stage)

Denied Credit 0.0063 0.0068 0.0082 0.0073 0.0060*** *** *** ** **

Bankrupt 0.0068 0.0071 0.0074 0.0079 0.0072*** *** *** *** **

Late Payments 0.0036 0.0037 0.0041 0.0027 0.0023** ** ***

Home Owner -0.0018 -0.0023 -0.0055 -0.0035 -0.0021**

Education X X X XDisabled X X XCollege-age kids X X XNormal income X XCounty UR X XCounty income X XPredicted wage X

Gicheva and Thompson Student Loans and Financial Stability October 26, 2013 20 / 26

Page 25: The Effects of Student Loans on Long-Term Household ...edpolicy.umich.edu/...thompson-presentation-effects-student-loans.pdf · The E ects of Student Loans on Long-Term Household

Results Main Sample

Results (First Stage)

F statistic of excluded instrument 57.47

Average loans per FTE 0.00317***

Female 0.381College degree 3.601***Masters degree 9.074***Doctorate 16.15***White 3.239**Black 4.280**Hispanic -0.121Age -0.680**Age sq 0.00679*Any college-age kids 2.356***Disabled -0.357County relative per-capita income 1.560*County unemployment rate 0.350***Predicted earnings 0.912***

Gicheva and Thompson Student Loans and Financial Stability October 26, 2013 21 / 26

Page 26: The Effects of Student Loans on Long-Term Household ...edpolicy.umich.edu/...thompson-presentation-effects-student-loans.pdf · The E ects of Student Loans on Long-Term Household

Results Main Sample

Results (First Stage)

F statistic of excluded instrument 57.47

Average loans per FTE 0.00317***Female 0.381College degree 3.601***Masters degree 9.074***Doctorate 16.15***White 3.239**Black 4.280**Hispanic -0.121Age -0.680**Age sq 0.00679*Any college-age kids 2.356***Disabled -0.357County relative per-capita income 1.560*County unemployment rate 0.350***Predicted earnings 0.912***

Gicheva and Thompson Student Loans and Financial Stability October 26, 2013 21 / 26

Page 27: The Effects of Student Loans on Long-Term Household ...edpolicy.umich.edu/...thompson-presentation-effects-student-loans.pdf · The E ects of Student Loans on Long-Term Household

Results Non-completers

Student Debt and College Completion

22 TRENDS IN STUDENT AID 2012 trends.collegeboard.org

See Figure 2011_9A and Figure 2011_9B Distribution of Cumulative Debt Among 2009 Degree Completers and Non-Completers.

Student Debt

FIGURE 11CAmount Borrowed by Students Who First Enrolled in 2003-04 and Left Without Completing a Degree or Certificate by 2009, by Institutional Sector and Length of Enrollment (with Percentages of Students in Each Sector Within Enrollment Category)

NOTE: Percentages in the parentheses on the vertical axis correspond to percentage of all students with the specified level of debt.

NOTE: In Figure 11C, institutional sector refers to first institution attended. Sectors do not sum to 100% because public less-than-two-year and private nonprofit two-year-or-less institutions are excluded. Percentages on the vertical axis refer to the percentage of all students leaving school without a degree who were enrolled for the specified number of months and to the sector breakdown within those categories. For example, 37% of the students who left without a degree were enrolled for more than 24 months. Of this group, 52% began their studies in two-year public colleges.

SOURCES: Beginning Postsecondary Students Longitudinal Study (BPS), 2009; calculations by the authors.

FIGURE 11BSector Last Attended, as of 2009, Among Bachelor’s Degree Recipients Who First Enrolled in 2003-04, by Total Amount Borrowed (and Overall Percentage of Students in Debt Category)

FIGURE 11AEducational Attainment by 2009 of Students First Enrolling in 2003-04, by Total Amount Borrowed (and Overall Percentage of Students in Debt Category)

Only 2% of students who first enrolled at a postsecondary institution in 2003‑04 had borrowed more than $50,000 by 2009. Over 40% did not borrow and another 25% borrowed $10,000 or less.

•Among students who either did not borrow or borrowed $10,000 or less, 43% had left college with no credential by 2009, while only 5% of those with more than $75,000 in debt were in this situation.

•Eighty-five percent of those with debt higher than $75,000 and 65% of those with debt between $50,000 and $75,000 had earned bachelor’s degrees by 2009.

•Among students who first enrolled in 2003-04 and earned bachelor’s degrees by 2009, 2% had last attended for-profit institutions. Among those who had borrowed more than $75,000, 18% had last attended for-profit institutions.

•Among students who first enrolled in 2003-04 and earned bachelor’s degrees by 2009, 65% had last attended public four-year institutions. Among those who had borrowed $75,000 or more, 28% had last attended an institution in this sector.

Cum

ulat

ive

Deb

t Lev

el

Percentage0% 20% 40% 60% 80% 100%

Cum

ulat

ive

Deb

t Lev

el

Percentage0% 20% 40% 60% 80% 100%

Did Not Borrow(43%)

$1 to $10,000(25%)

$10,001 to $20,000(16%)

$20,001 to $30,000(8%)

$30,001 to $50,000(5%)

$50,001 to $75,000(1%)

$75,001 or more(1%)

Total

Did Not Borrow(36%)

$1 to $25,000(42%)

$25,001 to $50,000(16%)

$50,001 to $75,000(3%)

$75,001 or More(2%)

Total

No Degree, Left Without Return

No Degree, Still Enrolled

Attained Certi�cate

AttainedAssociateDegree

Attained Bachelor’s Degree

For-Pro�t Four-Year

Private Nonpro�t Four-Year

Public Four-Year

26%

15%

42%

53%

58%

65%

85%

31%

1%

9%

9%

10%

10%

11%

9%

9%

9%

17%

7%

3%

2%

1%

0%

9%

14%

16%

17%

16%

15%

15%

10%

15%

5%

43%

43%

24%

18%

14%

10%

36%

71%

68%

54%

40%

28%

65%

28%

31%

40%

49%

54%

32%

1%

1%

6%

10%

18%

2%

Percentage0% 20% 40% 60% 80% 100%

Tota

l Mon

ths

Enro

lled

(and

Sec

tor) G

reat

er th

an

24 M

onth

s (3

7%)

13−2

4 M

onth

s(3

1%)

Up

to 1

2 M

onth

s(3

2%)

For-Pro�t (18%)

Private Nonpro�t Four-Year (4%)

Public Four-Year (10%)

Public Two-Year (65%)

Total

For-Pro�t (25%)

Private Nonpro�t Four-Year (6%)

Public Four-Year (14%)

Public Two-Year (52%)

Total

For-Pro�t (10%)

Private Nonpro�t Four-Year (12%)

Public Four-Year (25%)

Public Two-Year (52%)

Total

$20,001 or More$10,001−$20,000$1−$10,000Did Not Borrow

22%

42%

54%

87%

70%

7%

31%

37%

65%

44%

11%

27%

39%

59%

45%

66%

48%

46%

12%

27%

57%

41%

45%

30%

40%

18%

21%

23%

27%

24%

9%

10%1%

1%

3%

27%

15%

12%4%

12%

41%

28%

20%

10%

18%

4%

1%

0%

0%

1%

10%

13%6%

2%

5%

30%

24%

19%5%

13%

Source: Trends in Student Aid, 2012

Gicheva and Thompson Student Loans and Financial Stability October 26, 2013 22 / 26

Page 28: The Effects of Student Loans on Long-Term Household ...edpolicy.umich.edu/...thompson-presentation-effects-student-loans.pdf · The E ects of Student Loans on Long-Term Household

Results Non-completers

Results by College Completion Status

-0.14

-0.12

-0.10

-0.08

-0.06

-0.04

-0.02

0.00

0.02

0.04

0.06

Denied Credit Bankrupt Late Payments Home Owner

R or SP attend, drop out

R & SP attend, R and/or SP drop out

R or SP attend, complete

R & SP attend, complete

Gicheva and Thompson Student Loans and Financial Stability October 26, 2013 23 / 26

Page 29: The Effects of Student Loans on Long-Term Household ...edpolicy.umich.edu/...thompson-presentation-effects-student-loans.pdf · The E ects of Student Loans on Long-Term Household

Discussion

Summary of Findings

Student debt likely associated with decreased financial stability

I Additional $10,000 in student loans predicts 6 percentage pointincrease in probability of being denied credit over 5 year period and 7percentage point increase in probability of bankruptcy over 10 yearperiod

I Borrowers more likely to be late on bill payments

I Home ownership rates may also be affected

Debt management seems particularly hard for non-completers

Many confounding factors make establishing causality hard

Gicheva and Thompson Student Loans and Financial Stability October 26, 2013 24 / 26

Page 30: The Effects of Student Loans on Long-Term Household ...edpolicy.umich.edu/...thompson-presentation-effects-student-loans.pdf · The E ects of Student Loans on Long-Term Household

Discussion

Why Does Student Debt Lead to Observed FinancialDistress?

Impact on disposable income, combined with liquidity constraints

Debt-to-income ratio and mortgage availability

Changing aggregate labor market conditions?

I If continued decline in the returns to postsecondary educationcombined with policy-induced increase in borrowing

I Unlikely - evidence of increasing high school-college wage differential

Gicheva and Thompson Student Loans and Financial Stability October 26, 2013 25 / 26

Page 31: The Effects of Student Loans on Long-Term Household ...edpolicy.umich.edu/...thompson-presentation-effects-student-loans.pdf · The E ects of Student Loans on Long-Term Household

Discussion

Why Does Student Debt Lead to Observed FinancialDistress?

High uncertainty in ex post return to a degree

I Cannot tell if borrowers are making optimal choices ex ante

I Particularly bad for non-completers

I Focus on expected return or better safety net?

What is the correct counterfactual?

Gicheva and Thompson Student Loans and Financial Stability October 26, 2013 26 / 26