NowComment
Document: Invite Print Info
More
Properties/Access Dates Add to Document Analytics Report Fullscreen (hide top links) Download & Export Options Tag Document Embed in another web page
2-Pane Combined
Comments:
Full Summaries Sorted

Does Student Loan Debt COntribute to Racial Wealth Gaps? A Decomposition Analysis

There is evidence of a large and growing student debt burden over the last decade. Previous research has shown that the presence of student debt jeopardized the short‐term financial wealth of U.S. households during the Great Recession. We examine the effects of student loan use on the wealth of U.S. households post‐recession, using recent data from the 2013 and 2016 Survey of Consumer Finances. We find that mean 2016 wealth for households with no outstanding student debt is more than four times higher than households with student debt. We find that living in a household at the 15th, 30th, 50th, 70th, and 85th percentile of the wealth distribution with student debt is associated with an 80%, 49%, 37%, 35%, and 36% wealth loss compared with a similar household with no student debt. Our decomposition results suggest that student loan use can explain between 3% and 7% of the Black‐White wealth gap across the wealth distribution but is insignificant in explaining the Hispanic‐White wealth gap.

In the United States, the student loan system was designed to serve as an instrument for social mobility. However, over the last decade, there is evidence of a large and growing student debt burden. A record one‐in‐five households in the United States now owe student loan debt relative to one‐in‐ten in 1989, and 45% of this debt is owed by people 35 years and younger (Fry [19]). Student loan debt became the largest source of non‐mortgage debt owed by U.S. households in 2010, (Bricker et al. [ 7]). It amounts to an overwhelming $1.44 trillion in November 2018, almost five times the amount in 2004 (Federal Reserve Bank report [16]).[ 1] During the same time, there has been a significant change in the financing of higher education—with state and local funding covering a smaller share, and students and their families bearing a greater proportion of rising college tuition costs through student loans (College Board [10]).[ 2]

Previous research has shown that the presence of student loan debt jeopardized the short‐term financial health of U.S. households during the period of the Great Recession, 2007–2009, a period of great economic hardship using data from the Survey of Consumer Finances (SCF) (Elliott and Nam [14]). Since the period of the Great Recession was unusual because of adverse labor market and financial market conditions, it is necessary to examine this issue in different time periods.

This article provides a thorough analysis of the effects of student loan use on household financial health in the post‐recession time period for U.S. households. Specifically, we look at three questions: ( 1) What are the effects of student loan use on the financial health (net worth) of U.S. households during the post‐recession period between 2013 and 2016? ( 2) What role does student loan use play in explaining the Black‐White and Hispanic‐White wealth gaps in the post‐recession period? ( 3) To what extent does student loan use contribute toward explaining the racial wealth gaps across various socioeconomic groups?

To put our results into context, it is important to first understand the effects of student loan use on wealth (net worth) during the Great Recession. Findings from Elliott and Nam ([14]) suggested that median 2009 net worth for households with no outstanding student loan debt was almost three times higher than for households with outstanding student loan debt. Their findings also suggest that living in a household at the 15th, 30th, and 50th percentile of net worth in 2007 with outstanding student debt was associated with a 285%, 56%, and 54% loss in 2009 net worth compared with a similar household with no student debt. This paper goes beyond the previous study by examining the role of student loan use in contributing to the racial wealth gap for different types of households across the wealth distribution using a detailed decomposition analysis.

We found that mean 2013 net worth for households with no outstanding student debt is three and a half times higher than for households with outstanding student debt, $642,549 versus $184,524 while for 2016 it is more than four times higher, $865,956 versus $209,232.

Our results, based on the SCF between 2013 and 2016, quantified the effects of student loan use on net worth for various wealth quintiles of the U.S. population using median regression techniques. We found that living in a household at the 15th, 30th, 50th, 70th, and 85th percentile of the wealth distribution with student debt is associated with an 80%, 49%, 37%, 35%, and 36% wealth loss compared with a similar household with no student debt after we controlled for a host of demographic and employment‐related measures. Short‐term wealth losses associated with student debt during the post‐recession period are a significant proportion of the household total wealth but less than those incurred during the Great Recession. However, these losses continue to be significant and disproportionate for lower wealth quintiles as well as for Black and Hispanic households across the wealth distribution.

Further, we found that having a higher income, living in a household with a 4‐year college graduate, being older, being married, or having health insurance are all associated with an increase in net worth. Other explanatory variables like having technical/service‐related employment or not working, welfare use, and belonging to Black or Hispanic race are associated with a decrease in net worth.

Our decomposition analysis suggests that differences in student loan use account for 5% of the mean wealth gap between Black and White households. The extent to which student loan use contributes to the Black‐White wealth gaps varies considerably across the distribution with its effects being more pronounced for households at the median of the wealth distribution. We find that student loan use does not contribute to explaining the Hispanic‐White wealth gap across the wealth distribution. Due to the insurmountable rise of student debt in the last decade, examining the skewed effects of student loan use across different racial and socioeconomic groups contributes to our understanding of rising wealth inequality.

LITERATURE REVIEW

A few studies have evaluated various aspects of student loan use and its impact on the economic outcomes of U.S. households. These include effects of student loan debt on expected hourly wages (Minicozzi [31]; Daniels Jr. and Smythe [11]), career choices (Rothstein and Rouse [34]), homeownership rates (Mezza et al. [30]; Shand [37]; Xu et al. [39]), net worth and retirement savings (Elliott and Nam [14]; Hiltonsmith [21]; Rutledge, Sanzenbacher, and Vitagliano [35]), financial distress (Bricker and Thompson [ 6]), delinquencies, and repayment burdens (Akers [ 1]; Houle and Berger [22]).[ 3]

The Pew Research Center has published numerous reports about the financial health of young families on a regular basis. For instance, Fry ([19]) provides a comparison of younger U.S. households that owe student debt with similar households with no outstanding student loan balances. In general, a majority of young households have very modest amounts of wealth because it takes time to accumulate assets. Using 2010 SCF the study concludes that households headed by a young, college‐educated adult without student debt obligations have about seven times the typical net worth of similar households with student debt.

Hiltonsmith ([21]) uses 2010 SCF and other data sets to predict potential wealth loss of indebted households across the lifespan. The findings suggest that about $53,000 in education debt leads to a lifetime wealth loss of about $208,000 and that young indebted households have lower retirement savings along with lower home equity in comparison with debt‐free households of similar age group. The author also predicts that the $1 trillion in outstanding student loan debt will total in overall lifetime wealth loss of about $4 trillion for households with student loan debt. Rutledge, Sanzenbacher, and Vitagliano ([35]) find that graduates with student debt have much lower 401(k) assets by age 30 than those without debt.

Numerical studies conclude that student debt threatens balance sheets of households not only directly, by increasing household liabilities, but also by reducing college graduates' ability to build other assets. Some studies suggest that the value of assets may be a better gauge of overall financial health than net worth, because these focus on the results of human capital development, rather than counting student debt as a liability, as in net worth measures. Elliott, Grinstein‐Weiss, and Nam ([15]) use the 2007–2009 SCF panel to assess whether student loan debt is associated with total assets. Their findings indicate that median 2009 assets for households with no outstanding student loan debt were higher than those for households with outstanding student loan debt ($207,000 in 2009 vs. $174,000 in 2007). Quintile estimates reveal that households at the 25th, 50th and 75th percentile with outstanding student loan debt have 20% less in assets in 2009 than in 2007 when compared to a similar household with no student loan debt.

Bricker and Thompson ([ 6]) discuss trends in education debt between 1989 and 2010. They also use the 2007–2009 SCF panel to explore causal effects of student loan debt on household financial distress. Their findings suggest that households with student debt are more likely to be late in paying bills, experience denial in credit more often, and have high payment‐to‐income ratios.

According to the American Student Assistance's Life Delayed survey (see Lammers [27]), 62% of respondents said their student debt posed a hardship on their personal budget when combined with all other household spending. Specifically, 35% of respondents said they found it difficult to buy daily necessities because of their student loans; 52% said their debt affected their ability to make larger purchases such as a car; 62% said they have put off saving for retirement or other investments; and 55% indicated that student loan debt affected their decision or ability to purchase a home.

Previous studies have examined factors that contribute to racial wealth gaps over the long term. Thompson and Suarez ([38]) provide an update on racial wealth gaps. Their findings suggest that nearly most of the Hispanic‐White wealth gap at the mean and median of the wealth distribution can be accounted for by differences in demographic characteristics and educational attainment. However, Black‐White wealth gaps at the mean and median of the wealth distribution can be accounted for by differences in demographic characteristics and homeownership. Zhang and Feng ([40]) also find that homeownership plays a significant role in explaining wealth disparities by race, ethnicity, and education at the mean and bottom of the wealth distribution. Menchik and Jianakoplos ([29]) find that financial inheritances also contribute to Black‐White wealth gaps. Finally, McKernan et al. ([28]) find that young families and families of color experienced the largest percentage declines in wealth as a result of the Great Recession. Our study is a first in exploring the role of student loan use in contributing to the racial wealth gaps.

DATA

We use the SCF, a triennial cross‐sectional survey conducted by the Federal Reserve Board. The surveys provide comprehensive microeconomic level data on assets and liabilities, including student loans, of a nationally representative sample of U.S. households. SCF also includes detailed income and demographic level information for these families. In the SCF, relatively wealthy families are oversampled to represent the entire wealth distribution, which is an added advantage when using the survey data. This is useful since the distribution of wealth for the United States is more skewed than the distribution of income or earnings. For our analysis, we use most recent data, a pooled cross‐section from the 2013 and 2016 surveys, the most recent survey data available. The 2013 and 2016 SCF include approximately 6, 000 and 6,500 families, respectively.

We use data from survey respondents rather than heads of households since the SCF do not provide information on the race of the head of household. The additional advantage of utilizing information from survey respondents is that the survey respondent is "the economically dominant single individual or the financially most knowledgeable member of the economically dominant couple" (Kennickell [24]). We used the macro provided by the SCF to construct some of the variables for our analysis.

Our analysis mainly focuses on total wealth measured as net worth excluding student loan debt.[ 4] Since we wanted to examine the effects of student loan debt on net worth for 2013 and 2016, we added the student loan debt component to net worth, which is measured as total assets minus total liabilities. In SCF, intangible assets consist of the sum of savings, checking, money market accounts, certificates of deposit, stocks, bonds, mutual funds, 401(k) plans, pension plan balances, individual retirement accounts (IRAs), the cash value of whole life insurance policies. Tangible assets include real estate and cars, as well as loans against these assets. Liabilities consist of credit card balances and other consumer loans, including student loans. All dollar variables have been inflation adjusted to 2016 dollars.[ 5]

Net worth is transformed using the inverse hyperbolic sine (IHS), since this transformation allows us to estimate percentage change specifications for nonpositive net worth values. IHS has been used in several studies where changes in wealth are studied (Elliott and Nam [14]; Pence [33]; Thompson and Suarez [38]; Zhang and Feng [40]).[ 6] The transformation is described as follows:

θ − 1 sinh − 1 θ w = θ − 1 ln θ w + θ 2 w 2 + 1 1 / 2

where θ represents a scaling parameter and w represents net worth.

EMPIRICAL APPROACH

Quantile Regression

We use quantile regression techniques to examine how the presence of student debt affects different quantiles of the wealth distribution (Koenker and Bassett Jr. [25]). Generally, this technique can look at the effects on percentiles of the wealth distribution. The mean effect, while useful, may mask very different effects in different parts of the wealth distribution. We also look at the Least Absolute Deviation (LAD) estimator, which estimates the median regression.

IHS wealth it = β 0 + β 1 Student Loan Use it + β 2 X it + δ D 16 t + u it

Our dependent variable is the IHS of wealth for household i in year t. We include a number of covariates in our analysis. We include a year dummy, D 16 t , for 2016 omitting 2013. X it represents the demographics of the survey respondent. We use a set of dichotomous variables indicating student loan use, whether the respondent has a 4‐year or postgraduate degree, their marital status, welfare use, and health insurance coverage.[ 7] The key variable for analysis was our dichotomous variable for student loan use, which represents whether the survey respondent or anyone in their household owed any money or had any loans for educational expenses. We include real income in 2016 dollars, and categorical variables for age, occupational prestige, and race.[ 8] Lastly, we include interaction terms for select covariates.

For ease of interpretation, we converted IHS net worth back into dollar figures using the following transformation:

1 2 e θ y + e − θ y β X .

The above represents a marginal effect of a change in independent variable X on dollars of wealth w, where y = θ−1sinh−1(θw), θ is a scaling parameter for IHS transformation, and βX is a coefficient for variable X.[ 9] The IHS marginal effects depend on the chosen value of θ. The regression estimates in this study were based on a θ value of 0.00042, the optimal value estimated by the maximum likelihood method.

Decomposition Methods

To quantify how much group differences in student loan use contribute to the wealth gap between racial groups, we use the standard (Oaxaca [32]) and (Blinder [ 4]) (OB) decomposition method for the mean and then use an influence function regression decomposition technique developed by Firpo, Fortin, and Lemieux ([17]) for various quantiles.[ 10]

The racial wealth gap, such as between Black and White families, can be divided into a component that is observed by group differences in observed characteristics, △ ^ X μ , and a component that is unexplained by these observed differences, △ ^ S μ . The OB decomposition assumes a linear relationship between the dependent and independent variables. The contribution of differences in observed traits between groups (such as student loan use) to differences in the returns to those traits are separately identified.

The overall difference in mean wealth transformed by the IHS, △ ^ 0 μ = IHS ‾ W − IHS ‾ R , can be decomposed by the following:

△^Oμ=β^W0−β^R0+∑k=1KX‾Rkβ^Wk−β^Rk⏟△^SμUnexplained+∑k=1KX‾Wk−X‾Rkβ^Wk,⏟△^XμExplained

where W denotes the White group, R denotes the compared group, β ^ g 0 and β ^ gk denote the estimated intercept and slope coefficients, respectively, of the regression model for group g, where g denotes the White group, W, or compared group, R. The mean of observed characteristics, k, is denoted by X ‾ k . The first term in the equation, △ ^ S μ , is known as the wealth structure effect and is also referred to as the "unexplained" effect. The second element, △ ^ X μ , captures a composition effect, also referred to as the "explained" effect. In the above decomposition, the composition and structure effects (aggregate decomposition), and the contribution of each covariate to these effects (detailed decomposition) can be computed. The detailed decomposition involves subdividing both △ ^ S μ , the wealth structure effect, and △ ^ X μ , the composition effect, into the respective contributions of each covariate, △ ^ S , k μ and △ ^ X , k μ , for k = 1, ..., K.

We utilize the recentered influence function (RIF) technique developed by Firpo, Fortin, and Lemieux ([17]) to decompose the racial wealth gaps at the 15th, 30th, 50th, 70th, and 85th percentiles of the wealth distribution. The advantage of using the RIF technique is that it allows us to represent the influence of an observation on a distributional measure such as a quantile using the influence function. After the RIF regressions are estimated, the coefficients can be used to perform the detailed decomposition just like in the standard OB decomposition technique. We first estimate the RIF by computing the sample quantile, Q ^ τ , and estimating the density using kernel methods.

The RIF is given as follows:

RIF IHS Q τ = Q τ + τ ‐ 1 IHS ≤ Q τ f IHS Q τ

where Qτ refers to the population τ‐quantile of unconditional distribution of IHS, fIHS refers to the density of the marginal distribution of IHS, and 1{IHS ≤ Qτ} is an indicator function.

An equivalent of OB decomposition for any unconditional quantile is given as follows:

△ ^ O τ = X ‾ R γ ^ W , τ − γ ^ R , τ ⏟ △ ^ S τ Unexplained + X ‾ W − X ‾ R γ ^ W , τ ⏟ △ ^ X τ Explained

where

γ ^ g , τ = ∑ i ∈ G X i · X i T − 1 · ∑ i ∈ G RIF ^ IHS Q τ · X i , g = W , R .

RESULTS

Sample Characteristics

Our sample is a pooled cross section of SCF data from 2013 and 2016, both periods showing economic expansion, especially after the Great Recession of 2007–2009. Summary statistics by year are shown in Table . Mean education loan amount rose from $27,548 in 2013 to $32,797 in 2016 along with the average number of households with student loan debt, 24% in 2013 to 26% in 2016.[ 11] The median net worth also rose from $87,300 in 2013 to $104,470 in 2016. Further, approximately 39% percent of households had a head of household who had either a 4‐year college or postgraduate degree in 2013, while in 2016 this figure was a bit lower, at 34%. The median respondent's age was approximately 51 in 2013 and 52 in 2016. The median household income was $46,668 in 2013 and $52,657 in 2016. The distribution with respect to occupation and marital status remained similar across both survey years.

Summary Statistics by Year

2013 2016
Characteristics No. or Mean % or Median No. or Mean % or Median
Education loan use 28,799,044.00 23.50% 32,238,597 25.59%
Amount of loans for education $27,548.06 $15,000.00 $32,797.35 $17,000.00
Net worth $534,896.60 $87,300.00 $697,901.40 $104,470.00
Four‐year college graduate 47,518,268 38.78% 42,870,686 34.03%
Age 51.16 51 51.68 52
Income $86,596.13 $46,668.45 $102,252.00 $52,657.09
Profession
Managerial or professional 35,024,305 28.58% 35,732,728 28.36%
Technical services 24,869,145 20.30% 27,057,850 21.48%
Other 21,711,964 17.72% 21,472,627 17.04%
Not working 40,924,657 33.40% 41,718,495 33.11%
Married 70,027,299 57.15% 71,452,898 58.39%
Use of welfare 17,081,664 13.94% 18,077,117 14.35%
Race
White 85,882,792 70.09% 85,711,518 68.03%
Black 17,904,989 14.61% 19,973,733 15.85%
Hispanic 13,041,734 10.64% 14,282,169 11.34%
Other 5,700,555 4.65% 6,014,281 4.77%
Has health insurance 96,491,519 78.75% 109,675,006 87.06%
Observations 122,530,070 125,981,701

1 Note: Weighted data from the SCF survey are used. Net worth is adjusted for the level of student debt. Amount of loans for education are only for those with student loans.

About 79% percent of households were covered by some type of health insurance in 2013 and this increased significantly to 87% percent in 2016. This increase can be partially attributed to the Affordable Care Act. Racial demographics across both years represented the U.S. population very well. Table describes attitudes toward savings by year. We observe that around 15% of households report spending more than their income, around 30% report spending the same amount as their income, and just over 50% report that they save less than their income.[ 12]

Summary Statistics for Attitudes Toward Savings by Year

2013 2016
Characteristics No. % No. %
Savings behavior in past year
Spent more than income 18,545,778 15.14% 19,001,261 15.08%
Spent same as income 39,023,614 31.85% 37,208,200 29.53%
Spent less than income 64,960,679 53.02% 69,772,240 55.38%
Observations 122,530,070 125,981,701

  • 2 Note: Weighted data from the SCF survey are used. Net worth is adjusted for the level of student debt. Amount of loans for education are only for those with student loans.
  • 3 Adjusted for durables purchases/investments. Survey respondents have different reasons for saving, even though they may not be saving all of the time.

Sample Characteristics by Student Loan Use

Tables and provide summary statistics by distinguishing between households with student loans and those without student loans. Mean 2013 net worth for households with no outstanding student debt is three and a half times higher than for households with outstanding student debt, $642,549 versus $184,524 while for 2016 it is more than four times higher, $865,956 versus $209,232. Households with outstanding student loan debt have a higher share of head of households reporting having a 4‐year college degree or more than households without student loan debt, 49% versus 36% in 2013 and 41% versus 32% in 2016.

Summary Statistics for 2013

2013
Student Loans No Student Loans
Characteristics No. or Mean % or Median No. or Mean % or Median
Net worth $184,524.10 $40,400.00 $642,549.30 $114,200.00
Four‐year college graduate 14,233,546 49.42% 33,284,723 35.51%
Age 40.98 39 54.29 54
Income $68,576.75 $50,726.57 $92,132.62 $45,653.91
Profession
Managerial or professional 10,667,092 37.04% 24,357,212 25.99%
Technical services 7,538,297 26.18% 17,330,847 18.49%
Other 5,107,762 17.74% 16,604,202 17.71%
Not working 5,485,892 19.05% 35,438,764 37.81%
Married 17,549,285 60.94% 52,478,014 55.99%
Use of welfare 4,802,691 16.68% 12,278,973 13.10%
Race
White 18,655,205 64.78% 67,227,586 71.72%
Black 6,560,748 22.78% 11,344,242 12.10%
Hispanic 2,323,868 8.07% 10,717,866 11.43%
Other 1,259,223 4.37% 4,441,332 4.74%
Has health insurance 21,920,116 76.11% 74,571,402 79.56%
Observations 28,799,044 93,731,026

4 Note: Weighted data from the SCF survey are used. Net worth is adjusted for the level of student debt.

Summary Statistics for 2016

2016
Student Loans No Student Loans
Characteristics No. or Mean % or Median No. or Mean % or Median
Net worth $209,232.20 $49,120.00 $865,956.50 $140,210.00
Four‐year college graduate 13,103,392 40.65% 29,767,294 31.75%
Age 40.94 39 55.37 56
Income $79,325.45 $57,720.28 $110,136.50 $50,631.82
Profession
Managerial or professional 12,347,079 38.30% 23,385,648 24.95%
Technical services 9,340,364 28.97% 17,717,486 18.90%
Other 5,454,567 16.92% 16,018,060 17.09%
Not working 5,096,586 15.81% 36,621,909 39.07%
Married 19,271,005 59.78% 52,181,893 55.66%
Use of welfare 5,116,492 15.87% 12,960,625 13.83%
Race
White 20,643,609 64.03% 65,067,909 69.41%
Black 6,719,656 20.84% 13,254,076 14.14%
Hispanic 3,269,043 10.14% 11,013,126 11.75%
Other 1,606,289 4.98% 4,407,992 4.70%
Has health insurance 27,885,492 86.50% 81,789,514 87.25%
Observations 32,238,597 93,743,104

5 Note: Weighted data from the SCF survey are used. Net worth is adjusted for the level of student debt.

The average age of respondents who live in households with student loans is 41 which is much lower than the average age of mid‐50's of respondents who belong to households without student loans across both survey years. While the average household income for households with student loans is much lower than households without student loans, $68,577 versus $92,133 in 2013 and $79,325 versus $110,137 in 2016, median household income for households with student loans is higher than households without student loans, $50,727 versus $45,654 in 2013 and $57,720 versus $50,632 in 2016.

Tables and provide a description of attitudes towards savings by distinguishing between households with student loans and those without student loans. This disaggregated picture reveals that among households with student loans, 19% report spending more than their income.[ 13] These statistics indicate that education expenses play a significant role in the saving behavior of households.

Summary Statistics for Attitudes Toward Savings for 2013

2013
Student Loans No Student Loans
Characteristics No. % No. %
Savings behavior in past year
Spent more than income 5,566,299 19.33% 12,979,479 13.85%
Spent same as income 8,474,006 29.42% 30,549,608 32.59%
Spent less than income 14,758,739 51.25% 50,201,940 53.56%
Observations 28,799,044 93,731,026

  • 6 Note: Weighted data from the SCF survey are used. Net worth is adjusted for the level of student debt.
  • 7 Adjusted for durables purchases/investments.

Summary Statistics for Attitudes Toward Savings for 2016

2016
Student Loans No Student Loans
Characteristics No. % No. %
Savings behavior in past year
Spent more than income 5,711,381 17.72% 13,289,880 14.18%
Spent same as income 9,652,656 29.94% 27,555,544 29.39%
Spent less than income 16,874,560 52.34% 52,897,680 56.43%
Observations 32,238,597 93,743,104

  • 8 Note: Weighted data from the SCF survey are used. Net worth is adjusted for the level of student debt.
  • 9 Adjusted for durables purchases/investments.

Figure shows student loan debt as a share of income by household income quintiles. When we consider the resources available to households to meet their liabilities or debts, the relative burden of student loans is much greater for those at the lower end of the income distribution. In 2013 outstanding student debt was 39% of the household income for households in the bottom fifth quintile of the income distribution. By contrast, for households in the top fifth quintile of the income distribution, outstanding student debt was only 3.5% of household income. Clearly, this disaggregate statistics are masked if all households were grouped together. For all households, outstanding student debt was 7.5% of household income. For the second, third, and fourth quintiles, the student debt burden has increased, while there is a slight decrease of this burden for the top and bottom quintiles of the income distribution.

joca12271-fig-0001.jpg Student Loan Debt as a Share of Income by Household Income Groups Note: Weighted data from the SCF survey are used.

Results from Median Regression

We report median regression results from the 15th, 30th, 50th, 70th, and 85th quintiles of 2013 net worth in Table . The presence of student loans is an important determinant of net worth after holding all other factors constant. The association between student loans and all percentiles of 2013 net worth is strong and consistently negative. However, for households with student debt and a college degree, the net worth loss due to student debt is offset by the gains from a college degree across the wealth distribution.

Median Regression Results for Net Worth

15th Percentile 30th Percentile 50th Percentile 70th Percentile 85th Percentile
$4,205 $19,700 $87,300 $250,600 $624,210
Student loan use −3,358.85 −9,585.16 −32,595.53 −87,619.02 −223,063.14
(261.88) (754.06) (3,072.74) (8,728.79) (22,859.90)
Income 0.01 0.04 0.29 1.04 3.14
(0.00) (0.00) (0.02) (0.06) (0.17)
Four‐year college graduate 5,567.65 18,473.37 67,728.06 163,367.79 341,834.26
(327.76) (745.94) (2,874.62) (7,811.10) (24,153.13)
Age (Base: below 35)
35–44 2,972.37 15,737.70 64,517.28 156,412.78 388,646.99
(299.66) (1,075.71) (4,567.14) (11,377.78) (29,905.01)
45–54 4,176.64 22,884.37 112,544.58 302,350.73 718,997.80
(351.25) (1,090.71) (4,435.15) (12,361.58) (31,982.11)
55–64 6,176.79 30,840.02 148,319.52 391,919.68 967,188.25
(379.02) (1,351.31) (4,309.91) (14,296.70) (34,897.55)
65–74 8,026.31 41,283.04 177,829.32 465,649.01 1,093,660.21
(467.24) (1,440.78) (4,547.35) (14,225.39) (43,415.61)
Over 75 9,530.74 46,019.39 202,401.68 532,425.85 1,217,426.39
(583.59) (1,590.65) (5,642.09) (17,541.44) (43,884.41)
Occupational prestige
(Base: Managerial or Professional)
Technical/services 2,594.92 −7,623.87 −26,028.42 −70,947.27 −167,446.50
(422.19) (1,140.40) (3,952.88) (10,963.00) (29,937.04)
Other −2,691.70 −7,201.65 −21,637.99 −61,875.55 −160,605.89
(449.27) (1,418.60) (3,899.45) (10,151.76) (29,214.08)
Not working −4,245.12 −13,946.98 −46,718.42 −113,563.94 −200,097.96
(462.41) (1,422.44) (3,930.90) (12,351.19) (36,852.16)
Married 6,783.20 21,606.35 70,479.94 149,594.29 270,375.17
(226.88) (790.90) (3,165.00) (7,758.56) (22,047.68)
Use of welfare −6,852.61 −31,171.45 −125,852.54 −313,683.91 −639,030.70
(440.93) (1,236.59) (4,921.97) (13,125.83) (41,860.03)
Race (Base: White)
Black −3,351.91 −15,494.96 −60,836.42 −174,202.01 −395,843.53
(342.12) (1,299.69) (3,617.16) (10,089.55) (43,422.88)
Hispanic −5,675.47 −18,525.84 −61,207.42 −140,619.03 −232,732.01
(398.05) (1,572.56) (5,100.59) (11,376.32) (43,639.25)
Other −1,832.39 −3,534.18 −430.64 26,917.07 120,440.00
(502.27) (1,969.30) (5,683.86) (13,497.16) (39,027.85)
Has health insurance 4,214.79 15,493.00 54,139.50 115,567.00 243,782.93
(310.18) (1,087.91) (4,346.46) (9,784.85) (29,781.34)
Year 2016 961.07 2,536.87 10,041.25 35,881.04 94,738.43
(211.60) (727.67) (2,575.70) (6,560.25) (19,844.34)
Constant 2,352.35 26,703.39 192,311.82 782,519.04 2,357,918.13
(515.64) (1,520.49) (6,203.10) (13,080.77) (45,172.95)

  • 10 Note: Standard errors are bootstrapped with 999 replications and are adjusted for imputation uncertainty. Net worth is adjusted for the level of student debt. The coefficients are marginal effects evaluated at quantiles for net worth in 2013. Net worth is adjusted for the level of student debt. Standard errors are in parentheses. Population size in 2013 and 2016 are 122,530,070 and 125,981,701, respectively.
  • 11 *p < .1, **p < .05, ***p < .01.

Living in a household at the 15th percentile with outstanding student debt and 2013 net worth of $4,205 is associated with a $3,359 decrease in net worth (80% loss) compared with a similar household with no student debt. Living in a household at the 30th percentile with outstanding student debt and 2013 net worth of $19,700 is associated with a $9,585 decrease net worth (49% loss) compared with a similar household with no student debt. Living in a household at the 50th percentile with outstanding student debt and 2013 net worth of $87,300 is associated with a $32,596 decrease (37% loss) in net worth compared with a similar household with no student debt. Living in a household at the 70th percentile with outstanding student debt and 2013 net worth of $250,600 is associated with a $87,619 decrease (35% loss) in net worth compared with a similar household with no student debt. Living in a household at the 85th percentile with outstanding student debt and 2013 net worth of $624,210 is associated with a $223,063 decrease (36% loss) in net worth compared with a similar household with no student debt.

A higher income, living in a household with a 4‐year college graduate, being older, being married and having health insurance are all associated with an increase in net worth. Living in a household with a 4‐year college graduate is associated with higher net worth of $5,568 at the 15th percentile (132% gain), $18,473 at the 30th percentile (94% gain); $67,728 at the 50th percentile (78% gain); $163,368 at the 70th percentile (65% gain); and $341,834 at the 85th percentile (55% gain) compared with living in a household without a 4‐year college graduate and similar net worth levels. Similar descriptive statistics have been mentioned by Fry ([19]), where net worth for young, college‐educated households is analyzed.

Other explanatory variables such as having technical/service‐related employment or not working, welfare use, and belonging to Black or Hispanic race have a significant negative association with overall net worth.[ 14]

For instance, living in a household with health insurance is associated with higher net worth: $4,215 at the 15th percentile (100% gain); $15,943 at the 30th percentile (79% gain); $54,140 at the 50th percentile (62% gain); $115,567 at the 70th percentile (46% gain); and $243,783 at the 85th percentile (39% gain) compared with living in a household without health insurance and similar net worth levels.

The contrast between racial minorities and white households is stark. Black households experience a decrease in net worth: $3,352 at the 15th percentile (80% loss); $15,495 at the 30th percentile (79% loss); $60,836 at the 50th percentile (70% loss); $174,202 at the 70th percentile (70% loss); $395,844 at the 85th percentile (63% loss) compared with White households with similar levels of net worth.

Similarly, Hispanic households experience a decrease in net worth: $5,675 at the 15th percentile (135% loss); $18,526 at the 30th percentile (94% loss); $61,207 at the 50th percentile (70% loss); $140,619 at the 70th percentile (56% loss); $232,732 at the 85th percentile (37% loss) compared with White households with similar levels of net worth.

We report median regression results, including race and student loan use, interaction terms in Table . The interactions are significant and reveal that the partial effect of student loan use on net worth also depends on race. In our model, the difference in net worth for Black families with student debt versus White families with student debt is negative $549, at the 15th percentile households. Similarly, the difference in net worth for Hispanic families with student debt versus White families with student debt is negative $1,888, at the 15th percentile households. This difference continues to persist for households in the 30th, 50th, 70th, and 85th percentiles.

Median Regression Results for Net Worth with Interactions

15th Percentile 30th Percentile 50th Percentile 70th Percentile 85th Percentile
$4,205 $19,700 $87,300 $250,600 $624,210
Student loan use −4,337.73 −14,533.13 −46,028.98 −114,189.47 −254,767.53
(330.45) (893.02) (3,776.33) (9,794.10) (26,134.54)
Income 0.01 0.04 0.29 1.04 3.11
(0.00) (0.00) (0.02) (0.06) (0.17)
Four‐year college graduate 5,506.39 18,656.18 68,441.42 163,783.24 339,523.11
(315.12) (724.43) (2,778.43) (8,699.62) (26,064.85)
Age (Base: below 35)
35–44 2,840.48 15,625.72 66,596.73 157,040.35 386,252.86
(301.13) (1,060.78) (4,470.12) (11,995.05) (32,505.22)
45–54 4,077.59 23,695.53 113,499.37 299,967.59 711,177.82
(316.79) (1,022.45) (4,298.16) (13,740.97) (31,230.33)
55–64 6,022.95 30,800.70 146,378.41 391,781.60 961,733.69
(383.72) (1,333.11) (4,192.51) (15,338.55) (35,102.35)
65–74 7,983.06 40,933.00 178,230.64 463,190.59 1,085,084.89
(533.85) (1,656.80) (4,640.03) (14,010.45) (39,561.10)
over 75 9,515.72 45,714.03 203,213.97 531,353.23 1,211,894.70
(530.52) (1,866.95) (5,838.93) (17,422.95) (43,366.16)
Occupational prestige (Base: Managerial or Professional)
Technical/services −2,430.27 −7,579.60 −25,360.86 −69,928.14 −169,048.49
(426.86) (991.65) (3,521.75) (9,761.37) (31,978.07)
Other −2,684.54 −7,193.79 −22,184.75 −59,648.47 −150,860.95
(397.31) (1,402.96) (3,581.70) (10,352.92) (32,636.18)
Not working −4,094.48 −14,106.83 −47,045.48 −115,107.11 −192,318.74
(458.57) (1,463.67) (3,736.62) (11,906.68) (35,117.37)
Married 6,886.96 21,654.44 71,146.76 151,311.35 274,633.20
(256.68) (721.44) (3,012.12) (7,509.52) (21,697.49)
Use of welfare −6,706.35 −30,371.38 −127,438.16 −312,838.95 −638,540.64
(441.54) (1,203.49) (5,196.17) (12,827.32) (37,733.94)
Race (Base: White)
Black −4,442.64 −20,725.53 −76,681.12 −206,261.19 −464,927.00
(420.19) (1,406.47) (4,470.18) (10,942.10) (46,808.98)
Hispanic −6,531.60 −23,325.78 −75,477.17 −162,616.89 −284,523.67
(491.22) (1,623.13) (6,456.14) (14,215.33) (63,157.13)
Other −2,042.13 −4,819.58 −3,735.31 20,526.69 154,973.33
(517.06) (2,250.06) (5,966.59) (15,599.21) (61,862.91)
Has health insurance 4,134.44 14,946.20 50,726.19 113,275.29 243,634.61
(327.02) (1,168.07) (4,436.02) (10,260.14) (31,092.12)
Year 2016 983.28 2,706.95 10,997.17 35,445.66 90,648.87
(240.02) (679.40) (2,605.47) (6,842.37) (19,016.13)
Interaction terms (Base: White×Student loan use)
Black×Student loan use 3,893.15 15,607.23 40,997.09 84,317.81 206,023.12
(585.66) (1,684.86) (6,178.94) (17,542.12) (87,036.77)
Hispanic×Student loan use 4,643.44 21,820.85 59,415.04 94,101.99 136,070.01
(1,751.18) (2,806.29) (11,172.49) (39,424.07) (84,027.90)
Other×Student loan use 696.00 5,361.42 14,586.26 27,265.81 −101,079.62
(1,887.11) (4,270.61) (16,553.53) (40,520.92) (96,081.68)
Constant 2,611.51 28,243.12 197,361.75 790,159.52 2,370,285.24
(492.23) (1,724.27) (6,729.52) (15,574.69) (49,879.44)

  • 12 Note: Standard errors are bootstrapped with 999 replications and are adjusted for imputation uncertainty. Net worth is adjusted for the level of student debt. The coefficients are marginal effects evaluated at quantiles for net worth in 2013. Standard errors are in parentheses. Population size in 2013 and 2016 are 122,530,070 and 125,981,701, respectively.
  • 13 *p < .1, **p < .05, ***p < .01.

The difference in net worth for Black families with student debt versus White families with student debt is negative $5,118, $35,684, $121,943, $258,904 at the 30th, 50th, 70th, and 85th percentiles of households, respectively. Similarly, the difference in net worth for Hispanic families with student debt versus White families with student debt is negative $1,505, $16,062, $68,515, $148,454 at the 30th, 50th, 70th, and 85th percentiles of households, respectively.

There is however, an encouraging result with respect to student loan use within groups. For Hispanic households, student loan use is associated with positive net worth at the bottom of the wealth distribution to the median: $306; $7,288; $13,386. This suggests that student loan use might helping Hispanic families at the bottom to the median of the wealth distribution in building wealth. However, the association between student loan use and net worth becomes negative for Hispanic households with student debt at the 70th and 85th percentiles. For Black households, the association between student loan use and net worth is negative throughout the wealth distribution with the exception of households at the 30th percentile that get a $1,074 gain in net worth, which is again encouraging. These results suggest that student loan use affects different consumer groups differently.[ 15]

Does Student Debt Contribute to the Racial Wealth Gap?

The previous section provided some evidence of the association of student debt with wealth disparities across racial groups and the wealth distribution. In this section, we quantify how student debt contributes to the racial wealth gaps using the decomposition method described earlier.

Table displays the detailed decomposition of the racial wealth gap between Blacks and Whites as well as Hispanics and Whites due to the mean difference for each of the eight variables. The OB decomposition suggests that 67% of the mean Black‐White wealth gap and 68% of the mean Hispanic‐White wealth gap is accounted for by the variables in our regression. These results are in line with Zhang and Feng ([40]) and Thompson and Suarez ([38]), studies that focus on the role of homeownership and inheritances in explaining the racial wealth gap over a different time horizon, 1989–2013.

Decomposition of Mean Wealth Differentials

Black/White Hispanic/White
Reference group: White coef. Share Share
Unadjusted mean wealth gap 4,252.27 4,163.31
E[IHS

W

]–E[IHS

R

] 		(148.16) (178.29)
Total explained 2,839.98 67% 2,827.87 68%
(114.69) (117.50)
Total unexplained 1,412.29 33% 1,335.45 32%
(131.28) (162.67)
Explained: Composition Effects Attributable to
Student loan use 150.51 5% −27.16 −1%
(22.71) (16.95)
Income 1,445.45 51% 1,208.65 43%
(74.76) (75.39)
Four‐year college graduate 188.30 7% 270.54 10%
(23.83) (30.65)
Age 352.62 12% 736.00 26%
(47.48) (58.14)
Occupational prestige 32.69 1% 135.80 5%
(13.51) (30.34)
Married 127.44 4% −4.22 0%
(27.40) (8.69)
Use of welfare 461.84 16% 269.67 10%
(49.41) (37.88)
Has health insurance 81.12 3% 238.58 8%
(18.34) (45.20)

  • 14 Note: Estimates are survey weighted. Net worth is adjusted for the level of student debt. Income denotes the logarithm of income and age is collapsed into 5 groups to avoid omitted group bias. Robust standard errors are in parentheses.
  • 15 *p < .1, **p < .05, ***p < .01.
  • 16 The variable W denotes white and R denotes the comparison group.

Differences in income play a significant role, 51% in accounting for Black‐White wealth gaps and 43% in accounting for Hispanic‐White wealth gaps, respectively. Differences in student loan use contribute to 5% of the Black‐White wealth gap but does not significantly explain the Hispanic‐White wealth gap. Differences in the age of the survey respondent, occupational prestige, education, use of welfare, and having health insurance contribute significantly to the racial wealth gaps. Differences in being married contribute to the Black‐White wealth gap but not to the Hispanic‐White wealth gap.

Tables and provide a detailed decomposition using Firpo, Fortin, and Lemieux's ([17]) RIF regression techniques at the 15th, 30th, 50th, 70th, and 85th percentiles of the wealth distribution for Black‐White and Hispanic‐White wealth gaps, respectively. The extent to which income, student loan use, age, college education, marriage, and welfare use explains racial wealth gaps varies significantly throughout the wealth distribution. The importance of income, student loan use, college education, and age increases in explaining the Black‐White as we move from the bottom to the top of the wealth distribution. These results suggest that while student debt is contributing to the Black‐White wealth gap throughout the wealth distribution, its adverse effects are magnified for households at the median and top of the wealth distribution. However, the importance of welfare use, marriage, and having health insurance, and occupational prestige decreases in explaining the Black‐White wealth gap as we move from the bottom to the top of the wealth distribution. The results for Hispanic‐White wealth gaps differ from the Black‐White wealth gap. Student debt and marriage do not adequately explain the Hispanic‐White wealth gap. The contributions of other variables, however, remain similar to the results we obtain for explaining the Black‐White wealth gap. It is also important to note that the portion of the racial wealth gaps explained is greater at the bottom of the wealth distribution, consistent with Thompson and Suarez ([38]) and Zhang and Feng ([40]).

Decomposition of Wealth Differentials: Black vs. White

15th 30th 50th 70th 85th
Percentile Reference group: White coef. Share Share Share Share Share
Mean RIF gap 4,651 5,282 4,845 3,982 3,931
E[RIF

τ

(IHS

W

)]–E[RIF

τ

(IHS

R

)] 		(170.2) (188.6) (239.0) (186.20) (172.54)
Total explained 5,384 116% 5,058 96% 3,561 74% 2,639 66% 2,044 52%
(334.1) (284.5) (163.9) (120.56) (100.65)
Total unexplained −733.8 −16% 223.4 4% 1,283 26% 1,342 34% 1,887 48%
(363.4) (301.1) (225.5) (177.08) (160.65)
Explained: Composition Effects Attributable to
Student loan use 187.0 3% 207.1 4% 233.9 7% 168.63 6% 135.10 7%
(52.73) (44.99) (34.40) (26.69) (22.43)
Income 1,466 27% 1,777 35% 1,590 45% 1,664 63% 1,831 90%
(139.2) (129.6) (95.47) (102.00) (109.81)
Four‐year college graduate 236.7 4% 287.8 6% 298.3 8% 317.46 12% 231.75 11%
(54.32) (49.80) (38.67) (36.35) (29.35)
Age 477.3 9% 680.2 13% 533.2 15% 369.57 14% 294.89 14%
(85.17) (105.9) (73.09) (52.52) (39.95)
Occupational prestige 89.37 2% 102.1 2% 26.11 1% 13.26 1% 0.35 0%
(38.01) (32.70) (22.69) (20.27) (18.89)
Married 611.6 11% 444.4 9% 261.1 7% 56.16 2% −93.96 −5%
(95.19) (71.53) (49.63) (38.80) (32.82)
Use of welfare 2,064 38% 1,322 26% 485.6 14% 6.69 0% −319.78 −16%
(231.4) (146.6) (64.64) (33.75) (37.50)
Has health insurance 251.9 5% 237.2 5% 132.9 4% 42.95 2% −34.92 −2%
(61.75) (49.86) (28.22) (14.21) (11.56)

  • 17 Note: Standard errors are bootstrapped with 500 replications. Estimates are survey weighted. Net worth is adjusted for the level of student debt. Income denotes the logarithm of income and age is collapsed into 5 groups to avoid omitted group bias.
  • 18 *p < .1, **p < .05, ***p < .01.
  • 19 The variable W denotes white, R denotes the comparison group, and τ denotes the percentile.

Decomposition of Wealth Differentials: Hispanic vs. White

15th 30th 50th 70th 85th
Percentile Reference group: White coef. Share Share Share Share Share
Mean RIF gap 4,490 4,902 4,900 3,693 3,739
E[RIF

τ

(IHS

W

)]–E[RIF

τ

(IHS

R

)] 		(243.6) (268.0) (280.1) (231.4) (193.1)
Total explained 4,410 98% 4,947 101% 3,769 77% 2,973 81% 2,342 63%
(316.8) (262.4) (171.9) (124.09) (111.65)
Total unexplained 79.31 2% −44.84 −1% 1,131 23% 719.66 19% 1,397 37%
(350.2) (342.1) (290.8) (218.4) (199.1)
Explained: Composition Effects Attributable to
Student loan use −32.76 −1% −35.91 −1% −40.89 −1% −29.99 −1% −24.08 −1%
(22.66) (23.61) (25.96) (18.67) (15.11)
Income 1,226 28% 1,486 30% 1,329 35% 1,397 47% 1,536 66%
(128.9) (119.4) (93.45) (98.77) (111.71)
Four‐year college graduate 341.3 8% 415.6 8% 430.5 11% 459.85 15% 335.73 14%
(75.31) (70.11) (51.06) (48.56) (39.65)
Age 980.2 22% 1,443 29% 1,145 30% 791.62 27% 601.56 26%
(128.0) (136.5) (99.62) (65.65) (51.23)
Occupational prestige −27.86 −1% 186.2 4% 241.2 6% 225.42 8% 179.16 8%
(94.96) (79.62) (48.87) (38.47) (34.64)
Married −20.60 0% −15.19 0% −9.18 0% −2.21 0% 3.15 0%
(43.90) (31.64) (18.75) (4.79) (6.86)
Use of welfare 1,205 27% 771.4 16% 283.4 8% 3.98 0% −186.29 −8%
(168.8) (106.0) (45.26) (19.98) (26.42)
Has health insurance 739.4 17% 696.2 14% 390.0 10% 127.90 4% −103.97 −4%
(153.2) (116.2) (66.88) (39.25) (32.33)

  • 20 Note: Standard errors are bootstrapped with 500 replications. Estimates are survey weighted. Net worth is adjusted for the level of student debt. Income denotes the logarithm of income and age is collapsed into 5 groups to avoid omitted group bias.
  • 21 *p < .1, **p < .05, ***p < .01.
  • 22 The variable W denotes white, R denotes the comparison group, and τ denotes the percentile.

DISCUSSION, CONCLUSION, AND POLICY IMPLICATIONS

With student debt becoming the second highest category in household debt, there are concerns related to its impact on household wealth. Using the SCF, we find that the association between the presence of student loans and net worth is strong and consistently negative across the wealth distribution. Our results are in line with results of Elliott and Nam ([14]) who find that living in households at 15th, 30th, and 50th percentile with outstanding student loan debt and 2007 net worth is associated with significant decrease in 2009 net worth compared with similar debt‐free households. Our results, however, show a relatively smaller loss in net worth associated with student loan use across the wealth distribution perhaps, due to the continued improvement in households' economic well‐being in recent years relative to the years of the Great Recession of 2007–2009. Nevertheless, the presence of student loan debt is also associated with a disproportionate decrease in wealth for poor households.

Policymakers have also been concerned with the growing racial wealth gap and the factors that contribute to it. This article contributes to our understanding of the quantitative significance of student loan use in accounting for the racial wealth gaps in an empirically rigorous way. Student loan use contributes significantly in explaining the Black‐White wealth gap and its adverse effects are magnified for Black households at the median, 70th, and 85th percentile of the wealth distribution. Student loan use does not contribute to the Hispanic‐White wealth gap in the time period we consider. While our detailed decomposition analysis is useful to quantify the contribution of various factors in explaining racial wealth gaps, they do not seek to recover behavioral or causal relationships. Previous studies have found that Blacks are more likely than Whites or Hispanics to have student loan debt and finance their own education (Braga [ 5]), and that debt and default among Black college students is at alarming levels. Black college graduates default at five times the rate of white BA graduates (21% vs. 4%), and are more likely to default than White dropouts (Scott‐Clayton [36]). Grinstein‐Weiss et al. ([20]) suggest that low‐ and middle‐income Black and White students who face similar liquidity constraints and borrowing risks are at unequal risk of accumulating education debt. Further, there is evidence of a financial literacy gap with Whites having higher financial literacy scores than minorities despite financial literacy education (Al‐Bahrani, Weathers, and Patel [ 2]).

There are several mechanisms through which current and future policy can help more Americans reduce burdens associated with student debt: consumer‐driven oversight of the student loan industry; financial literacy and education; transparency and full disclosure about the type of student loans; various ways of repayment; grant, merit aid, and financial aid programs; and assistance with complaints related to student loans.

The Dodd‐Frank Wall Street Reform and Consumer Protection Act established a Student Loan Ombudsman within the Consumer Financial Protection Bureau (CFPB) in 2010. The ombudsman is responsible for compiling and analyzing data on student loan complaints annually and making appropriate recommendations to the Secretary of the Treasury, the Director of the Consumer Financial Protection Bureau, the Secretary of Education, and Congress. From July 2011 through August 2017, the CFPB handled over 50,700 private and federal student loan complaints, and about 9,800 debt collection complaints related to private or federal student loan debt (Chopra [ 9]). A continued consumer‐driven oversight of the student loan industry by federal agencies is one significant way to inform and shape public policy.[ 16]

Under the Fair and Accurate Credit Transactions Act of 2003, the Financial Literacy and Education Commission (FLEC) was established. The commission was tasked to develop a national financial education website, MyMoney.gov, and a national strategy on financial education. There are numerous recent initiatives from the FLEC that have resulted in useful public‐private partnerships.[ 17] The FLEC also suggests that starting financial literacy education and planning as early as kindergarten and continuing well beyond high school is useful. There are already some initiatives that help students make informed decisions about planning for, finding, and paying for college, such as the likelihood of finishing their degree and earnings in their field of study and budgeting under economic shocks (Avery and Turner [ 3]). However, there is a need for policymakers to examine if there is a systemic bias in the provision of financial literacy education (Al‐Bahrani, Weathers, and Patel [ 2]).

Finally, full transparency in colleges and universities about the type of student loan (federal or private), repayment amounts for each loan, repayment options, and clarification about the difference between grant aid and loans would be useful.[ 18] There are proposals that suggest replacing the current student loan system with an income‐based repayment system to address the current student loan repayment crisis (Dynarski and Kreisman [12]).

Our analysis provides useful indications of which factors are more important in explaining racial wealth gaps across the wealth distribution so that causal explanations or behavioral relationships can be explored in more detail. For instance, since our decomposition indicates that differences in student loan use account for a significant fraction of the Black‐White wealth gap across the wealth distribution, this suggests exploring in more detail how Black and White households choose to take out debt across the wealth distribution.

There are a few limitations to our study. Similar to the previous study by Elliott and Nam ([14]), we cannot conclude that a larger student debt necessarily causes households to have lower net worth. We tackle this issue by controlling for a host of factors considered important in predicting household net worth. Unobserved economic conditions may lead to effects on household net worth, and therefore, our findings should be treated as being suggestive of a relationship between outstanding student loan debt and net worth amount.

A second limitation is that we used a pooled cross‐section from SCF 2013 and 2016. This allowed us to provide comprehensive evaluation of relationship between student loan debt and U.S. households net worth for the expansionary period a few years after Great Recession for households with similar levels of net worth. Since the data are not final, especially with the sufficient time lapse between SCF surveys, we could not examine whether the same indebted households actually improved their financial situation in terms of net worth accumulation between 2013 and 2016 and whether the gap in net worth amounts between households with student loan debt and without increased or decreased.

Continuously increasing college costs and necessity of college education has made student loans one of the major factors affecting everyday life of many individuals and households in the United States. Our article underscores the growing importance of student loan debt in contributing to wealth disparities across racial groups and across the wealth distribution. Thus, this article signals the need for elevated policy discussions on the highly complex issue of educational finance, the efficacy of various student debt relief plans as well as the effectiveness of policies that aim to reduce inequality in the context of the growing and unbearable student debt burdens faced by young households and households of color.

Footnotes

1 By the time this article is published, the student loan debt would have crossed the $1.6 trillion dollar mark, Student Loan Debt Clock.

2 While federal and state funding of higher education is contingent on tax revenues, it also competes with other forms of public spending, and within education budgets such as preschool, elementary, and secondary education spending.

3 For a comprehensive review of literature on the effects of student debt on financial well‐being, see Elliott and Lewis ([13]).

4 According to (Kuhn and Rios‐Rull [26]), in SCF, "Wealth is the net worth of households. Our definition includes the value of financial and real assets of all kinds net of various kinds of debts. Specifically, the assets that we consider are the following: residences and other real estate; farms and all other businesses; checking accounts, certificates of deposit, and other banking accounts; IRA/Keogh accounts, money market accounts, mutual funds, bonds and stocks, cash and call money at the stock brokerage, and all annuities, trusts, and managed investment accounts; vehicles; the cash value of term life insurance policies and other policies; money owed to friends, relatives, businesses, and others; pension plans accumulated in accounts; and other assets. The debts that we consider are housing debts, such as mortgages, home equity, and HELOCs; other residential property debts, such as those derived from land contracts and vacation residences; credit card debts; installment loans; loans taken against pensions; loans taken against life insurance; margin loans and other miscellaneous debts."

5 Prices are adjusted using CPI for all urban consumers.

6 See Burbidge, Magee, and Robb ([8]) and Johnson ([23]) for detailed discussions on the use of the inverse hyperbolic sine transformation.

7 We follow the theoretical framework as specified in Elliott, Grinstein‐Weiss, and Nam ([15]) to address our first question.

8 According to Kuhn and Rios‐Rull ([26]), the SCF definition of income includes "both government and private transfers. Specifically, the sources of income that we consider are the following: wages and salaries; both positive and negative income from professional practices, businesses, and farm sources; interest income, dividends, gains or losses from the sale of stocks, bonds, and real estate; rent, trust income, and royalties from any other investments or business; unemployment and worker compensation; child support and alimony; Aid to Dependent Children, Aid to Families with Dependent Children, food stamps, and other forms of welfare and assistance; income from Social Security and other pensions, annuities, compensation for disabilities, and retirement programs; income from all other sources including settlements, prizes, scholarships and grants, inheritances, gifts, and so on. In other words, the notion of income that we use attempts to include all before‐tax income received during the year. It approximately corresponds to the payments to the factors of production owned by the household plus transfers. However, it does not include the income imputed from the services of some assets such as owner‐occupied housing."

9 As similarly described in Pence ([33]), w = θ − 1 sinh θy = 1 2 θ e θy − e − θy and dw dX = dw dy dy dX = 1 2 e θy + e − θy β X .

Notation for the decomposition methods follows Fortin, Lemieux, and Firpo ([18]).

We use the PLOAN6 variable from SCF which represents the Total value of aggregate loan balance by loan purpose: education and professional expenses, 2016 dollars.

Survey respondents also report the primary reason for saving, the top two reasons being retirement for around 30% of households and liquidity or the future for around 35% of households. Education is the primary reason for saving for around 8% of households. Other reasons include saving for purchases, saving for a home, family, investment, etc. While 4% of households report that they cannot save in 2013, this statistic falls to 0.83% by 2016.

About 10% of households cite education as the primary reason for saving in 2013, falling to 18% and 8% in 2016, respectively. Among households without student loans, 14% report spending more than their income and 8% cite education as the primary reason for saving in 2013, falling to 7% for education in 2016.

Welfare is an indicator variable, which takes on a value of one if anyone in the household has received income from TANF, SNAP (food stamps), or other forms of welfare or assistance such as SSI.

We also took a deeper look at the effect of student loan use on total assets across the assets distribution. Our results suggest that student loan use is associated with a loss in total assets across the assets distribution with its effects being asymmetric and adverse for the bottom of the asset distribution. We also found a negative association between student loan use and income throughout the income distribution. The result are asymmetric with the most adverse effects observed at the bottom of the income distribution.

For instance, CFPB sued Navient, the largest student loan servicer in the United States, which services more than $300 billion in federal and private student loans of more than 12 million borrowers. Navient was sued for creating obstacles to repayment by providing bad information, processing payments incorrectly, and failing to act when borrowers complained, per CFPB news 2017.

For instance, the Treasury and the Department of Education (ED) collaborated with H&R Block and Intuit, Inc. to incorporate federal student loan repayment plan information in tax preparation software and websites. More than 100,000 people learned about how they or their children could lower student loan payments. The Treasury also shared student loan repayment information in tax refund mailings to more than 20 million Americans. The FLEC has also launched an FAFSA mobile app in 2019 that helps consolidate all information on student aid at one place and has interactive tools for parents, students, providers of student loans. The app uses "nudging" to assist in repayment and in warning when taking out more debt than necessary. It also assists in student retention in college by immediately issuing small ticket loans for emergencies.

This is particularly important as private loans do not offer the same protections and repayment options as federal student loans.

Venoo Kakar ([email protected]) is an associate professor in the Department of Economics at College of Business, San Francisco State University, Gerald Eric Daniels Jr. ([email protected]) is an assistant professor in the Department of Economics at Howard University, and Olga Petrovska ([email protected]) is a graduate student of Economics at San Francisco State University. An earlier version of this article was titled "Student Loan Debt and Financial Health of U.S. Households." We would like to thank Editor Ronald P. Hill, three anonymous referees, Christopher L. Foote, Jeffrey P. Thompson, Philip Gundy for excellent research assistance, the Federal Reserve Bank of Boston's Research Department, International Association of Applied Econometrics 2018, Southern Economic Association 2018, 2019 Federal Reserve System Community Development Research conference participants for their insightful suggestions and the American Economic Association's Summer Economics Fellows Program. Any remaining errors are our own.

REFERENCES

Akers, Beth. 2014. How Much Is Too Much? Evidence on Financial Well‐Being and Student Loan Debt. Washington, DC : The American Enterprise Institute (AEI).

Al‐Bahrani, Abdullah, Jamie Weathers, and Darshak Patel. 2019. Racial Differences in the Returns to Financial Literacy Education. Journal of Consumer Affairs, 53 (2): 572 – 599.

Avery, Christopher and Sarah Turner. 2012. Student Loans: Do College Students Borrow Too Much–Or Not Enough? Journal of Economic Perspectives, 26 (1): 165 – 192.

Blinder, Alan S. 1973. Wage Discrimination: Reduced Form and Structural Estimates. The Journal of Human Resources, 8 (4): 436 – 455.

Braga, Breno. 2016. Racial and Ethnic Differences in Family Student Loan Debt. Washington, DC: Urban Institute.

Bricker, Jesse and Jeffrey Thompson. 2016. Does Education Loan Debt Influence Household Financial Distress? An Assessment Using the 2007–2009 Survey of Consumer Finances Panel. Contemporary Economic Policy, 34 (4): 660 – 677.

Bricker, Jesse, Lisa J. Dettling, Alice Henriques, Joanne W. Hsu, Lindsay Jacobs, Kevin B. Moore, Sarah Pack, John Sabelhaus, Jeffrey Thompson, and Richard A. Windle. 2017. Changes in US Family Finances from 2013 to 2016: Evidence from the Survey of Consumer Finances. FEDS Notes.

Burbidge, John B., Lonnie Magee, and A. Leslie Robb. 1988. Alternative Transformations to Handle Extreme Values of the Dependent Variable. Journal of the American Statistical Association, 83 (401): 123 – 127.

Chopra, Rohit. 2017. Annual Report of the CFPB Student Loan Ombudsman. US Consumer Financial Protection Bureau.

College Board, Trends. 2018. Trends in College Pricing. New York, NY: College Board.

Daniels Jr., Gerald E. and Andria Smythe. 2019. Student Debt and Labor Market Outcomes. AEA Papers and Proceedings, 109 : 171 – 175.

Dynarski, Susan and Daniel Kreisman. 2013. Loans for Educational Opportunity: Making Borrowing Work for Today's Students. The Hamilton Project Discussion Paper 5.

Elliott, William and Michal Lewis. 2015. Student Debt Effects on Financial Well‐Being: Research and Policy Implications. Journal of Economic Surveys, 29 (4): 614 – 636.

Elliott, William and IlSung Nam. 2013. Is Student Debt Jeopardizing the Short‐Term Financial Health of US Households? Federal Reserve Bank of St. Louis Review, 95 (5): 405–424.

Elliott, William, Michal Grinstein‐Weiss, and IlSung Nam 2013. Does Outstanding Student Debt Reduce Asset Accumulation? CSD Working Paper No. 13–32. St. Louis, MO: Washington University, Center for Social Development.

Federal Reserve Bank report, New York. 2018. Quarterly Report on Household Debt and Credit. Federal Reserve Bank of New York.

Firpo, Sergio, Nicole M. Fortin, and Thomas Lemieux. 2009. Unconditional Quantile Regressions. Econometrica, 77 (3): 953 – 973.

Fortin, Nicole M., Thomas Lemieux, and Sergio Firpo. 2011. Chapter 1 ‐ Decomposition Methods in Economics. In Handbook of Labor Economics, vol. 4 (1 – 102). San Diego, CA: Elsevier.

Fry, Richard. 2012. A Record One‐in‐Five Households Now Owe Student Loan Debt. Washington, DC: Pew Research Social & Democratic Trends.

Grinstein‐Weiss, Michal, Dana C. Perantie, Samuel H. Taylor, Shenyang Guo, and Ramesh Raghavan. 2016. Racial Disparities in Education Debt Burden among Low‐and Moderate‐Income Households. Children and Youth Services Review, 65 : 166 – 174.

Hiltonsmith, Robert. 2013. At What Cost? How Student Debt Reduces Lifetime Wealth. Dēmos.

Houle, Jason N. and Lawrence Berger. 2015. Is Student Loan Debt Discouraging Homeownership among Young Adults? Social Service Review, 89 (4): 589 – 621.

Johnson, Norman L. 1949. Systems of Frequency Curves Generated by Methods of Translation. Biometrika, 36 (1/2): 149 – 176.

Kennickell, Arthur B. 2010. Try, Try Again: Response and Nonresponse in the 2009 SCF Panel. In Proceedings of the Section on Survey Research Methods.

Koenker, Roger and Gilbert Bassett Jr. 1978. Regression Quantiles. Econometrica: Journal of the Econometric Society, 46 (1): 33 – 50.

Kuhn, Moritz and José‐Victor Rios‐Rull. 2016. 2013 Update on the US Earnings, Income, and Wealth Distributional Facts: A View from Macroeconomics. Quarterly Review, Federal Reserve Bank of Minneapolis, April : 1 – 75.

Lammers, Julie R. 2015. Life Delayed: The Impact of Student Debt on the Daily Lives of Young Americans. American Student Assistance.

McKernan, Signe‐Mary, Caroline Ratcliffe, Eugene Steuerle, and Sisi Zhang. 2014. Disparities in Wealth Accumulation and Loss from the Great Recession and Beyond. American Economic Review, 104 (5): 240 – 244.

Menchik, Paul L. and Nancy A. Jianakoplos. 1997. Black‐White Wealth Inequality: Is Inheritance the Reason? Economic Inquiry, 35 (2): 428 – 442.

Mezza, Alvaro, Daniel R. Ringo, Shane Sherland, and Kamila Sommer. 2016. On the Effect of Student Loans on Access to Homeownership. Finance and Economics Discussion Series 2016‐010.

Minicozzi, Alexandra. 2005. The Short Term Effect of Educational Debt on Job Decisions. Economics of Education Review, 24 (4): 417 – 430.

Oaxaca, Ronald. 1973. Male‐Female Wage Differentials in Urban Labor Markets. International Economic Review, 14 (3): 693 – 709.

Pence, Karen M. 2006. The Role of Wealth Transformations: An Application to Estimating the Effect of Tax Incentives on Saving. The BE Journal of Economic Analysis & Policy, 5 (1); doi.org/10.1515/1538-0645.1430.

Rothstein, Jesse and Cecilia E. Rouse. 2011. Constrained after College: Student Loans and Early‐Career Occupational Choices. Journal of Public Economics, 95 (1–2): 149 – 163.

Rutledge, Matthew S., Georey Sanzenbacher, and Francis M. Vitagliano. 2018. How Does Student Debt Affect Early‐Career Retirement Saving? Center for Retirement Research at Boston College Working Paper.

Scott‐Clayton, Judith. 2018. The Looming Student Loan Default Crisis Is Worse than We Thought. Evidence Speaks Reports. Center on Children and Families at Brookings 2.

Shand, Jennifer M. 2007. The Impact of Early‐Life Debt on the Homeownership Rates of Young Households: An Empirical Investigation. Federal Deposit Insurance Corporation Center for Financial Research.

Thompson, Jeffrey P. and Gustavo Suarez. 2015. Exploring the Racial Wealth Gap Using the Survey of Consumer Finances. Finance and Economics Discussion Series 2015‐076.

Xu, Yilan, Carrie Johnson, Suzanne Bartholomae, Barbara O'Neill, and Michael S. Gutter. 2015. Homeownership among Millennials: The Deferred American Dream? Family and Consumer Sciences Research Journal, 44 (2): 201 – 212.

Zhang, Sisi and Shuaizhang Feng. 2017. Understanding the Unequal Post‐Great Recession Wealth Recovery for American Families. The BE Journal of Economic Analysis & Policy, 17 (4); doi.org/10.1515/bejeap-2016-0229.

~~~~~~~~

By Venoo Kakar; Gerald Eric Daniels and Olga Petrovska

Reported by Author; Author; Author

Copyright of Journal of Consumer Affairs is the property of Wiley-Blackwell and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.

DMU Timestamp: January 21, 2020 02:52





Image
0 comments, 0 areas
add area
add comment
  • Join an existing conversation — click the “Reply” button of the appropriate right pane comment
  • Start a new conversation on an existing area — Double click on the existing highlighted area or its comment balloon
  • Define and comment on a new area — Draw a box around the desired area by clicking and then holding + dragging your mouse
Start a new conversation about the image as a whole (no specific area) — click here
change display
hide comment areas
hide comment balloons
help
Video
add comment
  at current time
  at a different time
on whole video
help

Quickstart: Commenting and Sharing

How to Comment
  • Click icons on the left to see existing comments.
  • Desktop/Laptop: double-click any text, highlight a section of an image, or add a comment while a video is playing to start a new conversation.
    Tablet/Phone: single click then click on the "Start One" link (look right or below).
  • Click "Reply" on a comment to join the conversation.
How to Share Documents
  1. "Upload" a new document.
  2. "Invite" others to it.

Logging in, please wait... Blue_on_grey_spinner