Volume 48, Issue 1 p. 5-21
Original Article
Full Access

What Factors Are Important in Aversion to Education Debt?

HanNa Lim,

HanNa Lim

Kansas State University

Search for more papers by this author
Jae Min Lee,

Corresponding Author

Jae Min Lee

Minnesota State University

Authors’ Note: HanNa Lim, PhD, is an Assistant Professor in the School of Family Studies and Human Services, College of Human Ecology, Kansas State University. Jae Min Lee, PhD, Assistant Professor, Department of Family Consumer Science, College of Allied Health and Nursing, Minnesota State University. Kyoung Tae Kim, PhD, is an Assistant Professor in the Department of Consumer Sciences, University of Alabama. Please address correspondence to Jae Min Lee, Department of Family Consumer Science, College of Allied Health and Nursing, Minnesota State University, Mankato, 102 Wiecking Center, Mankato, MN 56001; e-mail: jae-min.lee@mnsu.edu.Search for more papers by this author
Kyoung Tae Kim,

Kyoung Tae Kim

University of Alabama

Search for more papers by this author
First published: 09 September 2019

Abstract

This study examined the factors related to attitude toward education debt. Based on the conceptual framework of the multidimensional nature of attitude toward debt, empirical models were constructed with variables focused on debt utility and debt fear. Using the 2016 Survey of Consumer Finances dataset, this study found that debt utility variables, such as having a higher educational level and holding a student loan balance, were negatively associated with an aversion to education debt. However, debt fear variables, such as unfavorable attitudes about debt in general and not having applied for credit, were positively associated with aversion to education debt. Subgroup analyses across different income and educational levels showed that the negative relationship between education level and aversion to education debt was more evident in the high-income group and that most of the debt utility and debt fear variables were statistically significant only in the lower education level. These results suggest that aspects of debt utility and fear of debt contribute to attitudes about education debt, but there is inconsistency across socioeconomic status.

Because of the increasing rates of college attendance and increasing expenses for education, borrowing money for higher education has become commonplace for many Americans. According to a report published by Sallie Mae (2018), more than four in ten households with undergraduate students borrowed money for college expenses during the 2016–2017 academic year. Many people rely on education debt to benefit from higher education, because higher education is regarded as a necessary investment that can result in premium earnings and job security. For example, using the Current Population Survey, Torpey (2018) found that the median weekly earnings for college graduates were $1,173 in 2017, which is approximately 65% higher than that of high school graduates. Furthermore, the unemployment rate was 2.5% for college graduates, which is lower than the 4.6% rate for high school graduates (Torpey, 2018). Even considering the costs of a college degree, average college graduates can typically recoup the cost by age 40 (Daly & Bengali, 2014).

However, debt aversion, defined as an unfavorable attitude toward debt, may prevent individuals with limited financial resources or information from attending and completing college. Some media outlets have reported that younger generations such as Millennials and Gen Z are debt-averse (New, 2014; Rosen, 2017). Compared to earlier cohorts, a larger percentage of Millennials have a college degree (Munnell & Hou, 2018), and many of them obtained student loans to afford college expenses (Kurz, Li, & Vine, 2018). However, they entered the labor market during the period when the economy suffered from the dot-com bubble and the Great Recession (Munnell & Hou, 2018). This experience may lead young adults to discount the benefits of having a college degree and borrowing to obtain one and lead them to become debt-averse. Pew Research Center has revealed that the share of young adults holding debt of any kind and the amount of debt they are holding decreased significantly more than those of older adults between 2007 and 2010 (Fry, 2013).

Also, debt aversion in specific demographic groups, such as low-income households and first-generation college students, was found to narrow their options and act as a barrier to a college education (Burdman, 2005). Cunningham and Santiago (2008) found a pattern in which students from higher-income quartile households were more likely to borrow than were their lower-income counterparts. Among those who have substantial college expenses, 59% of the students from the highest income quartile borrowed money, while only 43% from the lowest income quartile did so. Also, the authors found evidence that those who were debt-averse were more likely to work while in college and to leave college without a degree compared to those who were willing to borrow. Despite the critical effect that education debt attitudes have on borrowing decisions and subsequent enrollment and completion decisions, little research has been conducted to examine attitudes toward education debt.

This study focuses on the aversion to borrowing for education and incorporates two recent discussions on debt attitude. First, the study followed the recent findings on the multidimensional nature of attitudes toward debt (Haultain, Kemp, & Chernyshenko, 2010). We examined the relationships among debt utility, debt fear, and attitudes toward education debt. Second, the study incorporated discussions on the disparity in specific groups’ attitudes toward debt and behavior (Burdman, 2005; Callender & Jackson, 2005). Analyses were conducted separately for subgroups by income and educational levels to examine how the relationships between debt utility and fear and attitudes toward education debt differ across the subgroups.

Review of Literature

Attitude toward Debt

Previous studies have used different terms for attitudes toward debt. Some researchers have used neutral terms, such as credit attitude (Chien & DeVaney, 2001; Godwin, 1997; Pattarin & Cosma, 2012) and debt attitude (Almenberg, Lusardi, Säve-Söderbergh, & Vestman, 2018; Davies & Lea, 1995; Haultain et al., 2010). Other researchers have used terms with more emotional valence, such as debt/loan aversion (Boatman, Evans, & Soliz, 2017; Burdman, 2005; Callender & Jackson, 2005; Cunningham & Santiago, 2008) and debt tolerance (George, Hansen, & Routzahn, 2018). Because this study explicitly addressed unfavorable attitudes toward education debt, we used the term education debt aversion.

Researchers have assumed that the attitude to debt is unidimensional, such as tolerant versus intolerant and negative versus positive (Haultain et al., 2010). However, recent findings have provided the support for a multidimensional structure of debt attitudes (Boatman et al., 2017; Haultain et al., 2010; Pattarin & Cosma, 2012). Using the responses of New Zealand students on attitude toward debt, Haultain et al. (2010) conducted factor analyses and extracted two factors for debt attitudes, “fear of debt” and “debt utility.” Fear of debt measures the extent to which people are averse to debt, and debt utility measures the extent to which they regard it as useful. The authors concluded that having no fear of debt and perceiving it to have considerable utility can predict debt decisions.

Pattarin and Cosma (2012) tested credit attitude overall among Italian households and the three components of attitudes: cognitive, emotional, and behavioral. Credit users had significantly more favorable attitudes overall and in all three attitude components compared to nonusers. Boatman et al. (2017) also discussed loan aversion's multidimensional structure and emphasized that it varies by context. They assessed loan aversion in three distinct ways: (i) respondents’ attitudes toward borrowing, (ii) respondents’ perception that it is acceptable to borrow money to pay for education and (iii) respondents’ preferences between cash and financial aid packages with grants and loans. In samples of high school seniors, community college students, and adults, the authors found that the three measures of loan aversion did not strongly correlate, providing evidence of debt attitudes’ multidimensional structure.

Factors Related to Attitude toward Debt

While previous studies have focused more heavily on investigating attitude toward education debt as the factor related to the decision to borrow, in this section, we review studies that have focused on debt attitude itself. As market constraints, such as high interest rates, may restrict borrowing behavior, examining the attitude toward borrowing can capture the real preference for borrowing. Using Swedish data, Almenberg et al. (2018) found the intergenerational transmission of debt attitudes. Those respondents with parents who felt uncomfortable with the debt itself were more likely to report that they were uncomfortable having debt. Further, the authors found that those with greater financial assets were more likely to have an unfavorable attitude toward debt, while those who were willing to take risks were more likely to have a favorable attitude toward debt. Gender, education, and financial literacy were not significant.

George et al. (2018) investigated gender, marital status, macroeconomic events, and personal economic experiences as the determinants of debt tolerance, specifically, for living expenses and luxury purchases. They found that personal economic experiences, such as unemployment and being late on payments, were significantly associated with tolerance toward borrowing for living expenses, while never having married was significantly associated with tolerance toward borrowing for luxury purchases.

Despite household debt's pervasiveness, researchers have found some evidence of debt aversions in specific socioeconomic groups. In particular, students from low-income households, students enrolled in community college, first-generation college students, and Hispanic students were debt-averse (Burdman, 2005). Using data from prospective students in the United Kingdom, Callender and Jackson (2005) used two different measures of debt attitudes: the general level of debt aversion and the perceived costs and benefits of attending university. They found that students from low-income households showed a higher level of debt aversion but perceived less benefit of attending university compared to those from medium- or higher-income households. Further, even after controlling for income, those with debt aversion and negative perceptions of the costs and benefits of the university were less likely to apply to universities. The authors concluded that debt aversion in the United Kingdom is an issue related to socioeconomic status.

Boatman et al. (2017) did not find evidence of attitudes that were more loan-averse among low-income respondents in three samples of U.S. high school seniors, community college students, and adults without higher education experience. Instead, they found that Hispanics were more averse to borrowing for education, while students whose parents attended college were less averse to doing so. Further, Haultain et al. (2010) found that those whose fathers and mothers attended university and had more savings were less fearful of debt. However, individuals from high socioeconomic groups concerning income, education, and occupation and those who had higher levels of debt felt that debt had greater utility.

Conceptual Framework

We developed the framework for this study based on the research on debt attitude's multidimensional structure (Haultain et al., 2010). We focused on two critical factors related to education debt attitudes—debt utility and fear of debt. We expect that those who experienced the value of education and perceived its importance are less likely to show an aversion to education debt. In economics, education is regarded as a form of investment in human capital (Bryant & Zick, 2005) due to the monetary and nonmonetary returns from education (Oreopoulos & Salvanes, 2011), and receiving an education and borrowing funds to pay for it may lead to positive attitudes toward education debt. However, individuals might assess debt utility differently depending on their personal experience and following cost and benefit analysis. We expect those who had negative experiences with borrowing and those who are averse to borrowing itself are more likely to show an aversion to education debt.

Specifically, we expect that respondents with higher levels of education, whose parents have higher levels of education, and respondents who have an education loan and have a saving motive for education will be less likely to exhibit aversion to education debt. Conversely, respondents who have experienced credit constraints and have an unfavorable attitude toward using credit in general will be more likely to report aversion to education debt. Further, we expect that the relationships between those factors and aversion to education debt to vary depending on socioeconomic status such as household income and educational attainment. See Figure 1.

Details are in the caption following the image
Framework of this study.

Methodology

Dataset and Sample Selection

We used the 2016 Survey of Consumer Finances (SCF) released by the Federal Reserve Board. The SCF dataset has been released triennially since 1983 and provides rich information on U.S. households’ financial status, particularly their balance sheet information. Further, some variables, such as parents’ financial knowledge and educational attainment, have been newly added to the 2016 SCF. The survey unit in the SCF is the primary economic unit (PEU), but we used households instead of following previous SCF studies. Because we are interested in the education debt attitudes in the general population, not in a specific demographic, the analytic sample in this study includes all 6,248 households in the 2016 SCF.

Dependent Variable

The dependent variable is a binary indicator of having an aversion to education debt based on how respondents feel about borrowing to finance educational expenses. The dependent variable was coded 1 if respondents reported that they do not feel it is all right for someone like himself or herself to borrow money for education expenses, and 0 otherwise.

Independent Variables

Variables related to debt utility

In this study, debt utility variables include four education-related variables, as follows: (i) respondent's education, (ii) parents’ education, (iii) holding a student loan and (iv) saving motive for education. The respondents’ highest level of educational attainment was measured as less than high school, high school diploma, some college, bachelor's degree, and post-bachelor's degree. The parents’ level of education was measured as a binary indicator of whether or not their highest level of education is a bachelor's degree or higher. We used higher education level for the respondents’ mother and father so that if either or both had a bachelor's degree or higher, it was coded 1, and if neither had a bachelor's degree or higher, it was coded 0. Student loan holders were measured as those holding an existing student loan balance. A saving motive for education was created based on whether or not the respondent reported that education was an important saving goal.

Variables related to fear of debt

Variables that measure fear of debt include two debt-related variables: credit experience and general attitude toward using credit. Following Lyons’ (2003) study on credit access, if the household had been turned down for credit in the past 12 months or was discouraged from applying for credit, it was defined as credit-constrained. However, those who reapplied for credit and obtained the loan amount desired were not considered credit-constrained. While Lyons (2003) used the information from the 1983 to 1998 SCF data that indicated whether the household had been turned down for credit in the past five years, we used turndown experience during the previous 12 months, which was available in the 2016 SCF. In addition, we split households without credit constraint into two groups, those who did not apply for credit and those who applied for and obtained credit. Therefore, three categories were used to measure credit experience: no application, no credit constraint, and credit constraint. The general attitude toward using credit was measured using the question of whether the respondent thinks buying items by borrowing or using credit is a good or a bad idea in general. Three response categories were available—favorable, neutral, and unfavorable.

Financial variables

Household income was measured as the total income received during 2015 before taxes and other deductions, and net worth was calculated based on numerous variables in assets and debts’ components. To capture the possible nonlinearity of the relationship between financial variables and the dependent variable, we used the natural logs of income and net worth after transforming nonpositive values to 0.01.

Behavioral variables

We created a continuous variable of objective financial knowledge based on the responses to three personal finance questions (i.e., stock, compounding, and inflation) that were developed by Lusardi and Mitchell (2006). The number of correct answers was summed and ranged from 0 to 3. Subjective financial knowledge was measured based on the respondents’ answer to the following question: “On a scale from zero to ten, where zero is not at all knowledgeable about personal finance and ten is very knowledgeable about personal finance, what number would you be on the scale?” Risk tolerance was measured in a way similar to subjective financial knowledge and ranged from 0 (not at all willing to take financial risks) to 10 (very willing to take risks). The risk variable also allowed respondents to choose −1 which meant “not at all willing to take financial risks.” Responses of −1 were recoded to zero. Uncertainty about next year's income was coded 1 if a respondent reported not having a good idea about next year's income, and 0 otherwise. A usual saver was measured based on whether the respondent usually: (i) set aside money for savings regularly, (ii) spent regular income and saved out of other income, or (iii) saved income from one family member while spending the other's income. If the respondent listed at least one of these responses as his/her saving habits, he/she was coded 1.

Demographic variables

The demographic variables included the respondents’ age, marital status and gender together, race/ethnicity, employment status, and presence of children. The coding for age was as follows: younger than 25, 25–34, 35–44, 45–54, 55–64, and 65 or older. Marital status and gender were coded as follows: married couple, single male, single female, or individual with a partner. Race/ethnicity was coded as White, Black, Hispanic, and Asian/others. Employment status was coded as follows: employee, self-employed, not working, or retired. The presence of children was coded as follows: whether or not the individuals had a child under age 18.

Analysis

This study investigated various factors related to the likelihood of having an aversion toward education debt. Given the binary dependent variable, the empirical model was specified using a logistic regression model. To isolate factors associated with education debt attitude across different income levels, we conducted analyses with subsamples categorized by income quartiles: low-income (bottom 25%), middle-income (middle 50%), and high-income (top 25%). We conducted additional analyses with subgroups categorized by the respondents’ educational attainment: low (less than a bachelor's degree) and high (bachelor's degree or above). For the multivariate analysis, the repeated-imputation inference (RII) technique was used to obtain more accurate estimates of variances, as Lindamood, Hanna, and Bi (2007) and Hanna, Kim, and Lindamood (2018) suggested.

Results

Descriptive Statistics

To compare respondents with and without education debt aversion, we reported descriptive statistics for two different samples in addition to the total sample. In the 2016 SCF, approximately 22% (1,368 of the total sample of 6,248) reported a negative attitude toward education debt while more than 78% (4,880 of the total sample of 6,248) reported a positive attitude toward education debt. This distribution contrasted with the distribution of general debt attitude. Only 25.9% of the total sample showed a favorable attitude, while 42.4% and 31.7%, respectively, showed neutral and negative attitudes about general credit use.

The two subgroups (i.e., those with a positive attitude toward education debt and those with a negative attitude) had similar demographic characteristics concerning marital status and race/ethnicity. In both subgroups, approximately half were married, and approximately one-fourth was single females. The majority in each group was White, followed by Black, Hispanic, and Asian/others. However, the mean age of those who were averse to education debt (58.5 years) was older than that of those who were not averse to education debt (49.0 years). Further, those with aversion to education debt had a higher proportion of “retired” (34%) and a lower proportion reported “have a child under age 18” (29.9%) compared to those with a positive attitude (17% and 43.9%, respectively).

The subgroups also showed different distributions concerning variables related to debt utility and fear of debt. Respondents who were averse to education debt had attained relatively lower levels of education, and their parents had as well. Approximately 56% of respondents with aversion to education debt completed more education than high school, and 21.2% had parents with a bachelor's degree, while 68.8% and 30%, respectively, of those with a positive attitude toward education debt completed more than high school and had parents with a bachelor's degree. Moreover, a lower proportion of education debt-averse respondents had an outstanding balance on their education loans (8.7%) and listed education as a saving motive (4.5%), compared to nonaverse respondents (15.8% and 7.9%), respectively. A lower proportion of the former had experienced credit constraints (11.3%) and had a favorable attitude toward using credit (16.7%) compared to the latter (18.4% and 28.5%, respectively).

With respect to financial and behavioral variables, those with a negative attitude toward education debt had a lower mean household income ($99,760) but they had a higher mean household net worth ($818,735), compared to those with a positive attitude to education debt ($102,951 for income and $653,360 for net worth), respectively. While both had similar levels of objective and subjective financial knowledge, the former showed lower risk tolerance and more uncertainty about income and had a lower proportion of usual savers compared to the latter. The level of average risk tolerance for the respondents with a negative attitude was 3.6, and it was 4.4 for those with a positive attitude toward education debt. Among those who were averse to education debt, approximately 30% were usually uncertain about their income next year, and 44.1% were usual savers, while among those who were not averse to education debt, 26.8% and 52.3% were uncertain about their income and were usual savers. See Table 1.

Table 1. Sample Characteristics
Variables Those with Negative Attitude toward Education Debt (= 1,368) Those with Positive Attitude toward Education Debt (= 4,880) All Households (= 6,248)
Variables related to debt utility
Education of respondent
Less than high school 16.8% 9.0% 10.7%
High school 26.9% 22.1% 23.2%
Some college 29.2% 30.2% 30.0%
Bachelor's degree 17.8% 23.4% 22.2%
Post-bachelor's degree 9.4% 15.2% 13.9%
Education of parent
Less than bachelor's degree 78.8% 70.0% 71.9%
Bachelor's degree 21.2% 30.0% 28.1%
Education loan holders
Yes 8.7% 26.1% 22.3%
No 91.3% 73.9% 77.7%
Saving motive for education
Yes 4.5% 7.9% 7.2%
No 95.5% 92.1% 92.8%
Variables related to fear of debt
Credit constraints
Applied for credit and approved 29.0% 41.0% 38.4%
Applied for credit but constrained 11.3% 18.4% 16.9%
Not applied 59.6% 40.5% 44.7%
General attitude toward using credit
 Favorable attitude 16.7% 28.5% 25.9%
Neutral attitude 36.1% 44.1% 42.4%
Negative attitude 47.2% 27.4% 31.7%
Financial variables
Mean (Median) income $99,759.8 ($41,518.1) $102,950.8 ($56,707.6) $102,252.0 ($52,657.1)
Mean (Median) net worth $818,734.7 ($97,300.0) $653,360.4 ($97,300.0) $689,575.9 ($97,300.0)
Behavioral variables
Objective financial knowledge of respondent (0–3) 2.1 (2.0) 2.2 (2.0) 2.2 (2.0)
Subjective financial knowledge of respondent (0–10) 7.2 (8.0) 7.3 (7.0) 7.3 (8.0)
Risk tolerance of respondent (0–10) 3.6 (3.0) 4.4 (5.0) 4.2 (5.0)
Uncertainty about income
Yes 29.6% 26.8% 27.4%
No 70.4% 73.2% 72.6%
Usual saver
Yes 44.1% 52.3% 50.5%
No 55.9% 47.7% 49.5%
Demographic variables
Age of respondent
Mean (median) age 58.5 years (61.0) 49.0 years (48.0) 51.0 years (51.0)
Younger than 25 6.9% 13.9% 12.4%
25–34 9.8% 19.4% 17.3%
35–44 12.5% 18.8% 17.4%
45–54 18.7% 19.7% 19.5%
55–64 23.0% 15.8% 17.4%
65 or older 29.1% 12.3% 16.0%
Marital status
Married 46.4 46.5 46.5
Single male 17.8 16.6 16.9
Single female 28.6 26.5 26.9
Partner 7.2 10.4 9.7
Race/ethnicity
White 69.5% 67.6% 68.0%
Black 14.0% 16.4% 15.9%
Hispanic 12.8% 10.9% 11.3%
Asian/others 3.8% 5.1% 4.8%
Employment status
Employee 37.9% 58.8% 54.3%
Self-employed 8.9% 9.6% 9.5%
Not working 19.2% 14.5% 15.6%
Retired 34.0% 17.0% 20.7%
Having a child under age 18
Yes 29.9% 43.9% 40.8%
 No 70.1% 56.1% 59.2%

NOTE

  • Weighted proportion. Reference categories in multivariate analyses are presented in bold.

Multivariate Results

Baseline results

Table 2 presents the results from a logistic regression on the likelihood of reporting an aversion toward education debt. Among variables related to debt utility, the respondent's education level, holding an education loan, and having a saving motive for education were related negatively to the likelihood of having an aversion to education debt. In particular, the odds of having an aversion toward education debt decreased in general as the respondent's education attainment increased. The odds of having an aversion to education debt among those with education loans were 0.49 times as low as the odds of those with no education loan. Further, the odds of having aversion to education debt among those who listed education as a saving motive were 0.72 times as low as the odds of those who did not. Concerning variables related to fear of debt, the odds of having an aversion to education debt on the part of those who had applied for credit were lower than for those who had not applied. Respondents who reported a negative or neutral attitude toward using credit in general were more likely to have a negative attitude toward education debt than were those with favorable attitudes.

Table 2. Logistic Regression Results on the Likelihood of Having an Aversion Toward Education Debt
Variable Coefficient SE p-Value Odds Ratio
Variables related to debt utility
Education of respondent (ref.: less than high school)
High school −0.3214 0.1228 .0088 0.7251
Some college −0.2742 0.1243 .0274 0.7602
Bachelor's degree −0.3607 0.1357 .0079 0.6972
Post-bachelor's degree −0.6228 0.1474 <.0001 0.5364
Education of parent (ref.: less than bachelor's degree)
Bachelor's degree 0.0076 0.0839 .9274 1.0077
Education loan holders (ref: No) −0.7109 0.1240 <.0001 0.4912
Saving motive for education (ref: No) −0.3352 0.1128 .0030 0.7152
Variables related to fear of debt
Credit experience (ref: not applied)
Applied for credit and approved −0.2571 0.0739 .0005 0.7733
Applied for credit but constrained −0.3143 0.1108 .0046 0.7303
General attitude toward using credit (ref.: Favorable attitude)
Neutral attitude 0.2699 0.0889 .0024 1.3099
Negative attitude 0.9861 0.0898 <.0001 2.6807
Financial variables
Log of income −0.0347 0.0158 .0284 0.9659
Log of net worth 0.0064 0.0087 .4656 1.0064
Behavioral variables
Objective financial knowledge of respondent −0.1274 0.0430 .0031 0.8804
Subjective financial knowledge of respondent 0.0368 0.0160 .0212 1.0375
Risk tolerance of respondent −0.0279 0.0128 .0300 0.9725
Uncertainty about income (ref: No) 0.1662 0.0768 .0305 1.1808
Usual saver (ref: No) −0.1242 0.0695 .0740 0.8832
Demographic variables
Age of respondent (ref.: Younger than 25)  
25–34 −0.0509 0.1639 .7559 0.9503
35–44 0.1453 0.1596 .3626 1.1564
45–54 0.2945 0.1557 .0585 1.3425
55–64 0.4572 0.1653 .0057 1.5797
65 or older 0.8125 0.1805 <.0001 2.2534
Marital status (ref.: Married)
Single male −0.1867 0.1031 .0702 0.8297
Single female −0.3751 0.0936 .0001 0.6872
Partner −0.1486 0.1326 .2623 0.8619
Race/ethnicity (ref.: White)
Black −0.0142 0.1095 .8965 0.9859
Hispanic 0.2874 0.1183 .0152 1.3329
Asian/others 0.2115 0.1563 .1759 1.2355
Employment status (ref.: Employee)
Self-employed 0.2983 0.0909 .0010 1.3475
Not working 0.2706 0.1927 .1602 1.3107
Retired 0.3940 0.1008 .0001 1.4829
Having a child under age 18 (ref: No) −0.2373 0.0826 .0041 0.7888
Intercept −0.964 0.2846 .0007  
Percentage concordant 72.4%      

NOTE

  • Unweighted results with RII technique.

Income was negatively associated with the likelihood of reporting an aversion to education debt, while net worth was not significantly associated with aversion to education debt. Among behavioral variables, objective and subjective financial knowledge showed the opposite relationship with aversion to education debt. Objective financial knowledge was negatively associated, but subjective financial knowledge was positively associated with the likelihood of having an aversion to education debt. In addition, risk tolerance was negatively related but uncertainty about income was positively related to the likelihood of having an aversion to education debt. Respondents aged 45 or older had a consistent pattern, that is, having a higher likelihood of reporting an aversion to education debt compared to those who were younger than 25. Single females and those with a child under age 18 had a lower likelihood of having an aversion to education debt, while Hispanic respondents, self-employed workers, and retired respondents had a higher likelihood of having an aversion to education debt. See Table 2.

Regression results across different income levels

Table 3 shows the results from three logistic regression analyses by income levels to isolate the associations between independent variables, including debt utility, fear of debt-related variables, and having an aversion to education debt across different income groups. Respondents’ education levels were negatively related to the likelihood of having an aversion to education debt, and the negative associations were more highly significant among the high-income group. Among the low- and middle-income groups, the odds of reporting an aversion toward education debt for education loan holders were 0.39 times (low-income) and 0.45 times (middle-income) as low as those without any education loan. Concerning variables related to fear of debt, those who had not applied for credit had a higher likelihood of reporting an aversion to education debt than did those who had applied for credit, although this association was significant only in the middle-income group. Respondents who reported a generally unfavorable attitude toward credit use had 2.41, 3.41, and 2.31 times higher odds, respectively, of having an aversion toward education debt than those with favorable attitudes across all income groups (low-, middle-, and high-income groups). See Table 3.

Table 3. Logistic Regression Results on the Likelihood of Having an Aversion Toward Education Debt
Variable Low-Income Group (Bottom 25%) Middle-Income Group (26%–75%) High-Income Group (Top 25%)
Coeff. SE Odds Ratio Coeff. SE Odds Ratio Coeff. SE Odds Ratio
Variables related to debt utility
Education of respondent (ref.: less than high school)
High school −0.3672* Significance level: *< .05, **< .01, ***< .001.
0.1857 0.6927 −0.2782 0.1991 0.7572 −0.5282 0.4495 0.5896
Some college 0.0448 0.2037 1.0459 −0.1970 0.1978 0.8212 −0.9297* Significance level: *< .05, **< .01, ***< .001.
0.4277 0.3947
Bachelor's degree −0.1618 0.2857 0.8506 −0.2864 0.2220 0.7510 −0.8727* Significance level: *< .05, **< .01, ***< .001.
0.4192 0.4178
Post-bachelor's degree 0.0269 0.3824 1.0273 −0.7096** Significance level: *< .05, **< .01, ***< .001.
0.2664 0.4918 −1.1506** Significance level: *< .05, **< .01, ***< .001.
0.4230 0.3164
Education of parent (ref.: less than bachelor's degree)
Bachelor's degree −0.1578 0.2055 0.8540 0.0322 0.1493 1.0328 0.0356 0.1153 1.0362
Education loan holders (ref: No) −0.9424** Significance level: *< .05, **< .01, ***< .001.
0.2937 0.3897 −0.8071 0.1850 0.4461 −0.4835* Significance level: *< .05, **< .01, ***< .001.
0.2249 0.6166
Saving motive for education (ref: No) −0.6125* Significance level: *< .05, **< .01, ***< .001.
0.2754 0.5420 −0.4571* Significance level: *< .05, **< .01, ***< .001.
0.2029 0.6331 −0.1628 0.1669 0.8497
Variables related to fear of debt
Credit experience (ref: not applied)
Applied for credit and approved −0.3326 0.1951 0.7171 −0.2894* Significance level: *< .05, **< .01, ***< .001.
0.1224 0.7487 −0.1603 0.1168 0.8519
Applied for credit but constrained −0.2347 0.2014 0.7908 −0.4681** Significance level: *< .05, **< .01, ***< .001.
0.1720 0.6262 −0.0515 0.2338 0.9498
General attitude toward using credit (ref.: favorable attitude)
Neutral attitude 0.0046 0.1952 1.0046 0.4440** Significance level: *< .05, **< .01, ***< .001.
0.1518 1.5589 0.2767 0.1432 1.3187
Negative attitude 0.8803*** Significance level: *< .05, **< .01, ***< .001.
0.1873 2.4117 1.2255*** Significance level: *< .05, **< .01, ***< .001.
0.1484 3.4057 0.8363*** Significance level: *< .05, **< .01, ***< .001.
0.1505 2.3079
Intercept −1.1389* Significance level: *< .05, **< .01, ***< .001.
0.4423   −1.3783*** Significance level: *< .05, **< .01, ***< .001.
0.391481   −1.7454 0.9166  
Other control variablesa aControl variables are the same as Table 2.
Yes     Yes     Yes    
Percentage concordant 74.9%     74.0%     70.3%    

NOTE

  • Unweighted results with RII technique.
  • aControl variables are the same as Table 2.
  • Significance level: *< .05, **< .01, ***< .001.

Regression results across different education levels

Similarly, we conducted two separate logistic regression analyses on aversion to education debt by different education levels. As shown in Table 4, most of the key variables were significant only in the low-education group. We found that education loan holders and those with a saving motive for education had 0.43 times and 0.61 times lower odds, respectively, of having an aversion to education debt than did the reference groups. With respect to variables related to fear of debt, respondents who had not applied for credit were more likely to have an aversion toward education debt than were those who had applied for credit. In addition, a generally unfavorable attitude toward credit was positively associated with the likelihood of having an aversion to education debt. These associations were significant in both groups. See Table 4.

Table 4. Logistic Regression Results on the Likelihood of Having an Aversion Toward Education Debt
Variable Low-Education Group (Respondents Less Than Bachelor's Degree) High-Education Group (Respondents with Bachelor's Degree or Above)
Coeff. SE Odds Ratio Coeff. SE Odds Ratio
Variables related to debt utility
Education of parent (ref.: less than bachelor's degree)
Bachelor's degree 0.0080 0.1252 1.0080 −0.0438 0.1076 0.9571
Education loan holders (ref: No) −0.8384*** Significance level: *< .05, **< .01, ***< .001.
0.1626 0.4324 −0.4530* Significance level: *< .05, **< .01, ***< .001.
0.1924 0.6357
Saving motive for education (ref: No) −0.4867** Significance level: *< .05, **< .01, ***< .001.
0.1638 0.6146 −0.1906 0.1596 0.8265
Variables related to fear of debt
Credit experience (ref: Not applied)
Applied for credit and approved −0.2397* Significance level: *< .05, **< .01, ***< .001.
0.1013 0.7869 −0.3002** Significance level: *< .05, **< .01, ***< .001.
0.1095 0.7407
Applied for credit but constrained −0.3242* Significance level: *< .05, **< .01, ***< .001.
0.1314 0.7231 −0.2631 0.2120 0.7687
General attitude toward using credit (ref.: favorable attitude)
Neutral attitude 0.2565* Significance level: *< .05, **< .01, ***< .001.
0.1160 1.2924 0.2544 0.1401 1.2897
Negative attitude 1.0073*** Significance level: *< .05, **< .01, ***< .001.
0.1146 2.7381 0.9279*** Significance level: *< .05, **< .01, ***< .001.
0.1443 2.5292
Intercept −1.1328 0.4021   −1.3627 0.4848  
Other control variablesa aControl variables the same as Table 2.
Yes     Yes    
Percentage concordant 72.4%     71.7%    

NOTE

  • Unweighted results with RII technique.
  • aControl variables the same as Table 2.
  • Significance level: *< .05, **< .01, ***< .001.

Discussion and Implications

This study examined the factors related to U.S. households’ attitudes toward education debt. Previous studies have usually focused on samples of college students or young student loan holders. However, this study analyzed a broader population, including those who had already completed their education. The analysis of the broad population allowed us to identify the underlying influences and interactions in a household's financial decisions as a collective decision unit. Lee, Kim, and Hong (2018) indicated that education debt decisions and the payment burden of household members could be pooled together for its financially interdependent feature. For example, parents or even grandparents often save for their children or grandchildren's higher education costs and obtain or cosign on loans for them. Increasing amounts of education debt have become an issue even among older Americans (Andriotis, 2019). Thus, this study included both household-level (e.g., net worth, income) and individual-level characteristics (e.g., parents’ education and debt aversion). In the future, studies on education debt decisions can examine these two-level characteristics (e.g., whether to take an education loan, how much it will be, who takes the loan, and the related attitude).

Individuals’ attitudes toward education loan debt show their perspective about higher education based on its expected costs and benefits, such as potential earning power and increased job security. For example, those who expect higher earnings to compensate for their education debt burden seem to be willing to assume student loan debt as a future investment. This study extends the existing discussions on debt attitude's multidimensional structure; individuals may have ambivalent attitudes toward debt, such as a simultaneous fear of debt and the perception that it is useful (Haultain et al., 2010). Most of the debt utility variables, such as respondents’ education, student loan holders, and saving motive for education, were negatively related to education debt aversion. However, variables that measure debt fear, such as a negative attitude toward using credit and not having applied for credit, were positively related to aversion to education debt.

Some variables did not support the hypothesized multidimensional structure of debt aversion. Among the debt utility variables, parental education (e.g., college graduate or not) was not a significant predictor, implying no direct effect of the previous generation's higher education experience on the next generation's attitude toward education debt. This result is inconsistent with previous studies that have reported a significant relationship between parents’ having a college degree and their children's education debt attitude (Boatman et al., 2017; Haultain et al., 2010). Debt fear from unexpected low returns in their education (Immerwahr & Johnson, 2009) can offset the debt utility from the positive earning of the previous generation with a college degree. This study also found that credit constraints had a negative effect on debt aversion. Compared to those who did not apply for loans, those who did were less likely to be averse to education debt, regardless of credit constraints. Among those who applied for loans, credit-constrained respondents were much less likely to be averse to education debt compared to nonconstrained respondents. These results appear to be puzzling, but they may reflect a difference in the two groups’ credit needs.

With respect to variables related to financial knowledge, it is notable that higher levels of objective financial knowledge were negatively related to aversion to education debt, while those of subjective financial knowledge had the converse effect. This opposite pattern between the two types of financial knowledge might be consistent with the findings of previous studies. For example, Grodsky and Jones (2007) indicated that disadvantaged parents tend to estimate their children's college costs imprecisely (e.g., overestimation) and make a biased decision for higher education. In this sense, the multidimensional debt attitude framework can encompass the underlying difference between those who value the utility of education debt and those who have a fear of debt. The difference can cause the former to assume a higher level of debt burden and the latter to voluntarily disregard a higher education's benefits.

This study's findings do not undervalue the importance of different debt decisions made by those in personal situations. Rather, imprecise assessment and perception about higher education's costs and benefits can be a potential risk factor that would reduce the opportunity to benefit from higher education (increasing earning power and upward social mobility). Thus, improving financial knowledge and providing accurate information about costs and benefits are essential to make a more informed education debt decision for those with and without debt aversion. If the education loan decision maker (student and family members) is more capable of using knowledge about benefits (expected earnings) and costs (short-term cost) and processing the relevant information (how the repayment plans work), the decision for a higher education can be changed (Slovic, Finucane, Peters, & MacGregor, 2004).

The burdens of student loan debt payments can affect debtors differently, depending on their life cycles (single, married, married with children, and retired) and characteristics (occupation, age, earnings, net worth, and the number of dependents). Some groups, including socio-economically vulnerable groups, may take longer to pay off or even fail to pay off the debt entirely (Looney & Yannelis, 2015). Being better equipped with how education loan repayment systems work can improve the decisions of the vulnerable group by expanding the number of available options. For example, there are alternative repayment plans of federal student loan programs depending on the debtor's income, family size, major, occupation, and so forth (such as the Income-Driven Repayment, Teacher Loan Forgiveness, and Public Service Loan Forgiveness). Loans of eligible debtors can even be forgiven, canceled, or discharged (The Office of Federal Student Aid, 2019).

This study's findings suggest a greater need for educating the socio-economically vulnerable groups by showing how personal experiences with education and knowledge about personal finances influence aversion to debt. For example, those with higher education levels and higher levels of household income were less likely to be averse to education debt. In particular, the negative relationship between education attainment and aversion to education debt was more prominent in the high-income group. However, those with lower levels of income and greater income uncertainty, and those who were older, Hispanic, self-employed, and retired were more averse to education debt than were their counterparts. Their potentially vague perceptions of debt or indefinite fear of debt can be an obstacle that prevents upward mobility or access to the increased social opportunities that higher education offers. Educators, researchers, and practitioners should pay greater attention to those vulnerable groups who are more likely to face the issue of lack of information (education debt repayment alternatives), knowledge (cost–benefit), and positive experience in completing higher education.

Our findings show that general aversion to credit was related to an aversion to education debt. Attitudes are not a static concept that never changes; instead, they are established and reshaped through several interactions within one's surroundings, including parents and educators. A thorough discussion about education loan debt cannot be made without an understanding of how people perceive it because the perception leads to behavior and the perception will direct the future issue of education loan debt. Findings on debt aversion can be examined in future research by incorporating potential changes in attitude before and after experiencing higher education or education loans. Studies can extend our discussion by analyzing different attitudes toward debt by generations (Millennials vs. Baby boomers) and intergenerational transmission of attitude toward debt.

Limitations

The study has some limitations. First, the 2016 SCF dataset was used to examine debt utility and the effect of fear on aversion to education debt. The use of longitudinal data or randomized experimental designs could allow future studies to examine any causal relationships between debt utility and fear and aversion to debt by addressing the potential endogeneity problem in the analysis.

Second, this study tested the multidimensional debt framework empirically by classifying education and debt variables into debt utility and fear by hypothesizing the positive and negative effect of each on debt aversion. However, some of our empirical findings were not well supported by the positive and negative effect of debt utility and fear framework. Although this study is based on a conceptual framework that has received recent attention in the literature on debt, our findings may not have empirically captured all aspects of debt utility and fear's structure. Thus, future studies can extend the results of this research by extending the relationship between more debt utility and fear variables and measuring each household's different levels of utility and fear.

Authors’ Contributions

Dr. Lim planned the study, analyzed the data, and wrote the initial draft. Dr. Lee helped write the initial draft and revised the manuscript. Dr. Kim helped with coding the data and writing the initial draft.

    The full text of this article hosted at iucr.org is unavailable due to technical difficulties.