Loan Aversion in Education: What We Know and What Remains To Be Learned

Transcription

1 Loan Aversion in Education: What We Know and What Remains To Be Learned BRENT EVANS PEABODY COLLEGE, VANDERBILT UNIVERSITY CUNY 4/21/17

2 Introduction Student loans are an important component of financing postsecondary education Media and literature primarily focus on over-borrowing However, some students under-borrow Avoiding borrowing may result in suboptimal human capital investment decisions May reduce likelihood of degree completion Some of this might be rationale We call these people loan averse Focus of a Lumina Grant funded project to better understand this phenomenon With my Peabody colleagues Angela Boatman & Adela Soliz

3 Background on student loans Federal and private Advantages to federal student loans Evidence that students don t exhaust federal borrowing before turning to private loans (Baum & Schwartz, 2013) Largest Source of Aid Only 10% of all undergraduate borrowers have a debt burden of $40,000 or more (College Board, 2016)

4 Student Loan Figures Debt of BA recipients Over Time Source: College Board

5 What We Know about Loan Aversion It appears to exist Qualitative evidence (Burdman, 2005; Cunningham & Santiago, 2008) International evidence (Caetano, Palacios, & Patrinos, 2011; Callender & Jackson, 2005; Palameta & Voyer, 2010) Domestic evidence for enrolled students (Goldrick-Rab & Kelchen, 2013) It might vary by student characteristics Cunningham & Santiago (2008) suggest Asian and Hispanic students are less likely to borrow

6 What We Would Like to Know about Loan Aversion How widespread is it? Does it vary across different groups and by individual characteristics? What causes it? Can we change it? Does it result in underinvestment in human capital? Our work attempts to address these questions, to varying degrees of success, using large scale quantitative data collection and analysis.

7 Limitations We observe mostly preferences not actual borrowing behavior We show there is a link between preferences and borrowing behavior with our community college sample for whom we actually observe borrowing, but it s not perfect. It would be great to observe borrowing behavior widely Requires longitudinal data tracking students into college We can still answer many interesting questions about borrowing preferences

8 Populations of Interest & Sample 3 different populations of interest High School Seniors Community College Students Adults in 20s and 30s without a college degree not enrolled in higher education Sample (~6,000) High school seniors from randomly selected diverse high schools in KY, TN, TX, and MA supplemented by some intentionally selected high school seniors in IL, and MI (captured >80% of seniors in most schools) CC students in TX, IL, MI, and TN Adults: Qualtrics survey using marketing lists to identify respondents 2 nd year study uses additional HS data from Jefferson Co. Public Schools

9 Survey Instrument Cross sectional Questions on Demographics Educational expectations Loan aversion measures Financial literacy and knowledge In person paper surveys for HS sample Online electronic survey for CC and Adult samples Pilot tested and reviewed by survey experts Conducted two separate randomized control trials using the survey

10 Descriptive Statistics for 3 Samples HS Sample Female White Black Hispanic Asian Multiple Races Age (0.56) Home Language Spanish Low-income Expect to get AA degree Expect to get BA degree Expect to get grad degree Parent Attended College N 1,648

11 Descriptive Statistics for 3 Samples HS Sample CC Sample Female White Black Hispanic Asian Multiple Races Age (0.56) (9.52) Home Language Spanish Low-income Expect to get AA degree Expect to get BA degree Expect to get grad degree Parent Attended College N 1,648 3,760

12 Descriptive Statistics for 3 Samples HS Sample CC Sample Adult Sample Female White Black Hispanic Asian Multiple Races Age (0.56) (9.52) (5.36) Home Language Spanish Low-income Expect to get AA degree Expect to get BA degree Expect to get grad degree Parent Attended College N 1,648 3,

13 Issues of Measurement Not much consensus in the literature on how to define or measure loan aversion General Attitudes: ask respondents their general opinions about borrowing money (Callender & Jackson, 2005) Attitudes about Borrowing Specifically: ask specifically about whether they think its ok to borrow for different purchases (Mortenson, 1988) Aid Packages: compare financial aid packages they would accept and observe if they avoid ones with loans (Palameta & Voyer, 2010); observe if they didn t accept loans offered to them (Goldrick-Rab & Kelchen,2013)

14 Our Loan Aversion Measures Attitudes (5 option Likert scale from strongly disagree to strongly agree) (Similar to Callender & Jackson, 2005) You should always save up first before buying something. Owing money is basically wrong. There is no excuse for borrowing money. Borrow for Education Do you think it s okay to borrow money to buy or pay for education? Aid Packages (Similar to Palameta & Voyer, 2010) Would you prefer $25 cash in one week or $1,000 grant once in college? Would you prefer $25 cash in one week or $1,000 grant + $1,000 loan once in college?

15 Paper 1 3 research questions To what extent is loan aversion present among high school students, community college students, and adults not enrolled in college? What is the relationship between different measures of loan aversion? Does loan aversion vary by individual characteristics? Descriptive analysis of survey responses

16 RQ1: Does loan aversion exist? You should always save up before you buy something. (Agree or Strongly Agree) Owing money is basically wrong. (Agree or Strongly Agree) There is no excuse for borrowing money. (Agree or Strongly Agree) Attitudes Scale (0.7960) HS Sample CC Sample Adult Sample (0.7406) (0.8420)

17 RQ1: Does loan aversion exist? You should always save up before you buy something. (Agree or Strongly Agree) Owing money is basically wrong. (Agree or Strongly Agree) There is no excuse for borrowing money. (Agree or Strongly Agree) Attitudes Scale (0.7960) Do you think it s okay to borrow for education? (No or I don t know) HS Sample CC Sample Adult Sample (0.7406) (0.8420)

18 RQ1: Does loan aversion exist? You should always save up before you buy something. (Agree or Strongly Agree) Owing money is basically wrong. (Agree or Strongly Agree) There is no excuse for borrowing money. (Agree or Strongly Agree) Attitudes Scale (0.7960) Do you think it s okay to borrow for education? (No or I don t know) HS Sample CC Sample Adult Sample (0.7406) (0.8420) Avoid Loan Packages N 1,648 3,

19 RQ2: Are these measures related? Not really Highest correlation between them is 0.25 between general borrowing attitudes and borrowing for education Suggests these are measuring different dimensions of loan aversion

20 RQ3: Does loan aversion vary by subgroup? Low Income Status: no difference Gender: Women are less likely to be loan averse on the attitudes and borrowing for education measures (effect sizes of ) Parents attended college: less loan averse Race: Relative to white students Black students more loan averse on fin aid packages Hispanic students more loan averse on all measures in HS sample Asian students are less loan averse on borrowing for education

21 Paper 1 Summary Loan aversion is wide spread, although different measures provide different estimates It s less prevalent in students actually enrolled in college suggesting it might deter some students from enrolling It varies by subgroup with women less loan averse and Hispanic respondents more loan averse Next two papers explore some financial correlates and behavioral causes of loan aversion

22 Paper 2 We demonstrated how loan aversion varies by demographics in the first paper Now we examine financial correlates of loan aversion, specifically Financial literacy Financial aid knowledge Credit market experience Uses same survey responses Divide CC sample into borrowers and non-borrowers Focus exclusively on the borrowing for education measure of loan aversion Although purely descriptive, these results shed light on potential causes of loan aversion

23 Summary of Results Financial literacy is associated with reduced loan aversion for non-borrower CC students (5 pp) and adults (9-10 pp) Knowledge about federal student loans is associated with reduced loan aversion Awareness of federal student loans for high school and community college students (3-8 pp) Awareness of subsidized interest for community college students (2 pp) Awareness of income based repayment for high school, community college non-borrowers, and adults (3-8 pp)

24 Summary of Results Prior experience with mortgage or student debt is associated with reduced loan aversion Mortgage for adults (5 pp) Student debt for high school and community college samples (6-10 pp) Having a payday or title loan is associated with increased loan aversion for CC non-borrowers (6 pp)

25 Paper 2 Conclusions Loan aversion is related to financial characteristics including financial literacy, knowledge about federal financial aid, and prior experience in the credit market. Not consistent across samples Unlike demographics, these financial characteristics can be changed. Increased knowledge may reduce loan aversion. Policy interventions can target knowledge gaps Financial literacy courses in high school and community colleges Providing information about the benefits of federal student loans and forms income based repayment

26 Paper 3 Considers behavioral effects explaining loan aversion Specifically focusing on framing and labeling effects Field (2009) demonstrated framing effects for law students Contacts framed as a grant instead of a loan induced enrollment and affected career choices We build directly off of Caetano, Palacios, & Patrinos (2011) Replicate their analysis in the U.S. among our three populations Extend to examine differential effects across demographics

27 Framing Component Suppose you need $10,000 to finance a one-year education program. In one year you will join the work force. How do you prefer to finance your education? (Choose one.) 60 monthly payments of $200. If in any month your income is below $2,000, then you only have to pay 10 percent of your income that month. 60 monthly payments equal to 10 percent of your income. If in any month your income is larger than $2,000, then you only have to pay $200 in that month. The two choices are financially equivalent, but one is framed as an income based repayment loan (option 1) and one is framed as an income share agreement/human capital contract (option 2).

28 Framing Results If there are no framing effects and students were perfectly rationale, we should see an equal split across these options, they should be indifferent between them. Chi-squared one sample goodness of fit test (% choosing ISA framing) High School Community College Adults Expected Percent Observed Percent Chi-squared p-value < For HS and CC students, clear preference for loan framing

29 Labeling Experiment Same question with randomly added Loan and Income Share Agreement labels to survey questions options: Loan: 60 monthly payments of $200. If in any month your income is below $2,000, then you only have to pay 10 percent of your income that month. Income Share Agreement: 60 monthly payments equal to 10 percent of your income. If in any month your income is larger than $2,000, then you only have to pay $200 in that month. Treatment effect of label: HS: 11 pp more likely to choose ISA, CC: 8 pp more likely to choose ISA, Adults: no effect

30 Heterogeneity Across Race and Risk Aversion Among High School Students Black and Hispanic students exhibit stronger framing and labeling effects (20-21 pp) Among Community College Students We see stronger labeling effects for Hispanic students than for white or black students We measured risk aversion attitudes on the survey and investigate whether the effects are stronger among risk averse students We see some suggestive evidence of larger labeling effects for more risk averse high school and community college students

31 Paper 4 Can we actually affect loan averse attitudes through an information intervention? Paper 2 identified that financial aid information is related to loan aversion Paper 3 suggested that risk may play a role in loan aversion Blocked clustered randomized controlled trial providing information to JCPS high school seniors Treatment: 5 minute video explaining federal loan system and advantages of it including income based repayment Control: 5 minute video explaining how to read a financial aid award letter Outcome measure: loan aversion as measured by our survey

32 Information Experiment Results Videos have large effect on aid knowledge ~30 pp affect on ability to answer loan or aid letter questions covered in the videos Treatment video reduces loan aversion on general borrowing attitudes scale (0.14 effect size) education specific loan aversion measure (30 percent reduction relative to control group mean of 16 percent, effect size of 0.40). Recently received data on college enrollment will enable us to see if this reduction in loan averse preferences are related to college enrollment.

33 What have we learned? How widespread is loan aversion? It s prevalent, although less so among enrolled community college students Does it vary across different groups and by individual characteristics? Yes, Hispanic students are particularly prone to be loan averse What causes it? Not conclusive, but evidence suggests it is related to knowledge and information and perhaps risk Behavioral effects such as framing and labeling are also contributing Can we change it? Information changes attitudes, which is encouraging Remains to be seen if information changes borrowing and enrollment behaviors Does it result in underinvestment in human capital? We don t know, we assume so, but this remains to be studied directly

34 Where do we go from here? Link loan averse attitudes to actual borrowing behavior We scratch the surface of this by dividing our community college sample into borrower/non-borrower Link loan aversion with postsecondary outcomes Working with TG on implementing loan aversion measures in a survey of Texas college students which would enable us to capture some of this Need additional data sources or ways to track students longitudinally

35 Policy Implications Information is important Increase financial literacy education Provide information on Income Based Repayment before borrowing decision not just at repayment decision Students don t like the framing and labeling Change the language used to describe borrowing in the literature Use Income Share Agreements Different measures aren t highly related, so when measuring loan aversion, do it multiple ways.

36 Questions/Discussion

37 Appendix slides 1-3: paper 2 measures: fin aid lit, fin aid knowledge, and credit market history 4: paper 2 descriptives on measures 5: paper 2 main table results 6 paper 3 risk aversion results

38 Financial Literacy Poor financial literacy has been linked to lack of retirement planning, lack of participation in the stock market, and poor borrowing behavior (Lusardi & Mitchell, 2008). The literature finds a positive association between improved financial knowledge and effective financial behaviors (Chen & Volpe, 1998; Borden, Lee, Serido, & Collins, 2007). Our financial literacy measures are two multiple choice questions that ask about interest rates and inflation.

39 Knowledge of the Federal Loan System Generally, federal student loans are better than private loans, but many people do not know the difference between the two forms of borrowing (CFPB, 2012). It is common for students to take out private loans without exhausting their eligibility for federal students loans (Baum & Schwartz, 2013). Knowledge of the federal loan system may improve borrowing decisions. Our knowledge measures include three true/false/yes/no questions about awareness of federal student loans, subsidized interest, and income based repayment.

40 Prior Experiences in the Credit Market Observed negative experiences with parental credit card debt is linked to negative perceptions of credit card usage (Joo, Grable, & Bagwell, 2003). Our measure of prior credit market experience asks if their household has any mortgage, student debt, or payday or title loans. Although we did not ask whether these experiences were positive or negative, we generally believe payday and title loans are negative experiences given their terms.

41 Financial Characteristics HS Sample CC Sample Adult Sample Financial Literacy Awareness of Federal Student Loans Awareness of Subsidized Interest Awareness of Income Based Repayment HH has Mortgage Debt HH has Student Loans HH has Payday/Title Loans N 834 3,

42 Relationship between Loan Aversion for Education and Financial Characteristics HS Sample CC Sample Adult Sample Financial Literacy *** *** (0.0245) (0.0052) (0.0278) Knowledge about Federal Student Loans Aware of Federal Student Loans ** ** (0.0268) (0.0125) (0.0374) Aware of Subsidized Interest * (0.0228) (0.0093) (0.0292) Aware of Income Based Repayment ** ** (0.0295) (0.0092) (0.0274) Prior Experience in the Credit Market Household has Mortgage * (0.0301) (0.0112) (0.0316) Household has Student Debt *** *** (0.0299) (0.0135) (0.0332) Household has Payday or Title Loan (0.0179) (0.0502) N 834 3,

43 Risk Seeking > Risk Averse Framing Effects High School Expected Percent Observed Percent Chi-squared p-value Labeling Effects High School Treatment ** (0.0903) (0.0954) (0.0425) (.01768) (0.0538) (0.0419) Control Mean Observations R-squared Framing Effects Community College Expected Percent Observed Percent Chi-squared p-value Labeling Effects Community College Treatment * *** (0.0545) (0.0693) (0.0590) (0.0851) (0.0357) (0.0458) Control Mean Observations R-squared