The Role of Institutional and State Aid Policies in Average Student Debt

Abstract

Rising college student debt levels have received considerable media coverage and have even prompted policy proposals that link rising student debt with tuition inflation. This article examines the role of state aid policies coupled with tuition and financial aid policy and academic outcomes in determining variation in average student debt. A focus solely on tuition as the culprit in rising student debt misses the significant role that state and institutional financial aid policies and student outcomes play in determining debt levels across higher education institutions. Specifically, colleges and universities being need-blind in admissions, meeting-full-need, limiting loans, and graduating students in high paying majors can have a larger impact on student debt levels than can the cost of attendance. Similarly, higher state-provided student aid significantly lowers average student debt at public universities.

Keywords

student debt higher education finance educational outcomes

Higher education institutions in the United States have come under increasing attack for graduating students with debt levels that are considered to be overly burdensome. Recent headlines highlight the fact that aggregate student debt has now surpassed total credit card expenditures in the United States. For example, a headline from an October 2011 USA Today issue reported, “Student Loan Debt Surpasses $1 Trillion” (Cauchon 2011). The article goes on to discuss that students borrowed more than $100 billion in the previous year alone. A 2012 report, Grading Student Loans, by the Federal Reserve Bank of New York points out that, of the 37 million borrowers who had outstanding student loan balances as of the third quarter of 2011, 14.4 percent, or about 5.4 million borrowers, have at least one past-due student loan account. Cumulatively, these past-due balances amount to $85 billion, or approximately 10 percent of the total outstanding student loan balance.

The College Board’s (2011) Trends in Student Aid reports that in the 2009–2010 academic year approximately 55 percent of public four-year college students who graduated from the institutions at which they began their studies graduated with some level of student debt. Their average cumulative level of borrowing was approximately $22,000. About two-thirds of those earning bachelor’s degrees from private nonprofit institutions had debt averaging $28,100. Borrowing in the Federal Stafford Loan Program (both subsidized and unsubsidized) among all students increased by 61 percent from 2000–2001 to 2010–2011, from $3,256 to $5,253 per FTE (full-time enrollment), in constant 2010 dollars (College Board 2011).

Clearly student borrowing and cumulative student debt levels have increased dramatically over the past decade. This rise in debt coupled with the stagnant employment market caused by the recession of 2007 to 2009 has led to increases in default rates on student debt. In 2000, the two-year student debt default rate was 5.9 percent. By 2010, the two-year default rate had risen to 9.1 percent.¹

The general angst surrounding student borrowing has prompted calls for political action to alleviate the onerous burden of student debt experienced by some borrowers and to hold institutions responsible for their graduates’ debt woes. For example, U.S. Representative Mike Pompeo, from Kansas, stated that he would like to make universities pay off at least part of defaulted loans owed by their students (Wenzl 2013). A recent Wall Street Journal op-ed (Reynolds 2013) called for legislation allowing student debt to be forgiven when declaring bankruptcy, an action that is not permitted under current bankruptcy laws, with the college or university that received the money being liable to the guarantors, that is, the U.S. taxpayers, for 10 to 20 percent of the balance. Similarly, President Obama, in a January 2012 speech at the University of Michigan, said that higher education has become an imperative for success in the United States, but the cost has grown unrealistic for too many families and the debt burden unbearable. This speech was followed in 2013 by the president’s proposal entitled, “A Better Bargain for the Middle Class: Making College More Affordable,” where he outlined a bill that could create a rating system designed to provide information to families on which institutions offer the best value. His plan calls for federal financial aid to ultimately be tied to this rating and for state funding of public colleges to be based on institutional performance and efforts to keep tuition increases in check.

The cacophony of discourse concerning the escalating cost of higher education and the rising levels of student debt is clearly beginning to have an impact on campuses. Colleges and universities are feeling tremendous pressure to control their tuitions and total costs or risk losing vital levels of federal support. A number of universities are undertaking various policies to reduce student indebtedness among their graduates. For example, according to a recent article by Craven (2013), the University of North Carolina at Chapel Hill has attempted to contain cost increases as a means of reducing student debt, while Albright College in Pennsylvania is attempting to lower student debt by meeting 100 percent of demonstrated need. The University of Michigan is decreasing student loan debt by holding tuition increases in check and simultaneously increasing financial aid.

All of this is occurring in an environment where state aid to public universities and grant aid to students are experiencing dramatic reductions (see Delaney, this volume). A Joint Economic Committee of the United States Congress report (2013) on the causes and consequences of increasing student debt reports that “state subsidies and grants, which were significant in the past, have declined, leaving students and their families to bear more of the financial burden … and increasing their reliance on federal financial aid, including grants and loans.” Cuts in state grants directly to universities and indirectly through grants to students may play an important role in determining variation in average student debt levels across institutions. Every state of the union (and Washington, D.C. and Puerto Rico) provides at least some direct grant aid to its undergraduate students. For example, according to a report by the National Association of State Student Grant and Aid Programs (2012), New Jersey, the state with the highest average need-based state grant per student, provided grants directly to students equal to approximately $976 per FTE undergraduate in 2010–2011. On the other hand, Georgia spent less than $4 per FTE undergraduate on need-based grant aid, but $1,895 per FTE undergraduate in total, primarily through its Hope Scholarship program, in 2010–2011.

Nonetheless, as Delaney (this volume) also describes, state grant aid to students has not kept pace with tuition inflation over the past decade. In the 2000–2001 academic year average state grant aid per FTE undergraduate represented approximately 14 percent of average tuition and fees at public four-year universities. By 2010–2011, state grant aid per FTE undergraduate covered only 8 percent of average tuition and fees at public four-year institutions, based on author calculations from data published in the 32nd and 42nd Annual Survey Report on State-Sponsored Student Financial Aid from the National Association of State Scholarship (2002, 2011) and Grant Programs and the 2012 Trends in College Pricing, from the College Board (2012). Variation in the level of state aid to postsecondary students and the mix of need-based versus merit-based aid may play a significant role in determining differences in average debt levels across institutions.

Institutional financial aid and admissions policies may also be important in determining student debt levels. Institutions that admit more affluent students should have lower levels of student borrowing than colleges that cater to the less well-to-do. Institutions have considerable discretion in establishing admissions and financial aid policies. They must weigh the benefits of various policies with the institutional resources available to fund these policies. For example, colleges and universities that are need-blind in admissions, a practice that is usually perceived to be advantageous for less affluent students, may have higher levels of average institutional grant and student debt, ceteris paribus, while institutions that pledge to meet the full demonstrated need of all enrolled students will likely have lower levels of student borrowing (if that need is met primarily with grant aid). On the other hand, institutions that “gap,” that is, they do not fully meet all of the demonstrated need of their students with financial aid (grants plus loans), would be expected to have higher levels of student borrowing, ceteris paribus.

While institutional prices, policies, and practices play an important role in determining average student debt, institutions cannot simply impose debt on an unwilling student body. Students make enrollment decisions based on price, value, and expected returns on their investment. Students who are more likely to graduate and who pursue majors in fields with greater earnings potential should be more willing to assume debt in pursuit of their degrees. As a result, one would expect institutions with higher graduation rates and a higher proportion of degrees awarded in more lucrative majors to have higher levels of average student debt, all else being equal.

The primary focus both in the media and the political arena has been on the role of tuition in determining student debt levels. Although tuition plays an important role, this article examines the role that state aid, admissions practices, financial aid policies, and student outcomes play in determining the level of student debt across four-year not-for-profit institutions, conditional on the cost of attendance.

Literature Review

There is a scarcity of rigorous empirical research that directly addresses state and institutional determinants of average levels of student debt. There is greater coverage of the importance of student loans on college enrollment and choice decisions and the implications of student debt on educational outcomes at the individual level. This brief literature review highlights the most recent and applicable studies of student loans and their implications.

The analysis in Macy and Terry (2007) most closely approximates the question addressed in this article. Examining data from the top 200 colleges and universities in the U.S. News and World Report, they conclude that tuition and fees are the single most important determinant of variation in average student debt levels across institutions. They also find that the percent of students with debt and the size of the institution are positively related to average institutional debt levels, while the endowment value, percent of classes with fifty or more students, alumni giving rate, and percent of Hispanic students are negatively related to average student debt levels across institutions. Their analysis, however, did not investigate the impact of admissions and financial aid policies or differences in state grant aid on variation in average student debt levels across institutions.

As institutional average debt levels are the product of both state and institutional policies and the enrollment decisions of students, a review of the literature on the influence of student loans on college choice decisions is warranted. The literature on the impact of student borrowing on college enrollment and specific college choice decisions is in general agreement on the positive influence that student borrowing has on enrollment decisions, at least for some students. For example, Buss, Parker, and Rivenburg (2004) utilize data from a set of selective liberal arts colleges and find that student loans increase the likelihood of enrollment, at least within this selective sector of higher education. Linsenmeier, Rosen, and Rouse (2006) take advantage of a unique natural experiment at a private, highly selective university that eliminated loans for all low-income students. They found that enrollment increased by a rather paltry 3 percentage points as a result of this policy change; the increase was not statistically significant. On the other hand, they found a more robust and statistically significant increase of 8 to 10 percentage points in the likelihood of enrollment among minority applicants. Examining the broader question of whether no-loan policies influence matriculation decisions, Waddell and Singell (2011) conclude that adopting a policy of no-loans (all grant aid) for low-income students leads to the matriculation of students with higher levels of financial need. The primary conclusion of these and most other studies is that reducing student loan levels has a positive, although not necessarily large, impact on the enrollment decisions of low-income students.

Another vein of the student debt literature examines the impact of student borrowing on measures of postgraduation outcomes. For example, Minicozzi (2005) finds that graduates with higher levels of student debt upon graduation are more likely to take higher paying first jobs, but jobs with lower wage growth. Tumen and Shulruf (2008) and Monks (2001) conclude that higher levels of student indebtedness do not discourage students from pursuing additional investments in education, while Luong (2010) reports that student borrowers have lower savings rates, less assets, less net worth, and are less likely to own their own home by age 29. Rothstein and Rouse (2011), once again utilizing a natural experiment at a single private highly selective (anonymous) university, find that higher levels of debt prompt graduates to choose higher paying jobs, and those with more debt are less likely to choose “public interest” jobs. This finding is consistent with a study by Field (2009), who finds that loans have a negative impact on choosing public service employment among a set of law school graduates from New York University. More recently, Avery and Turner (2012) frame higher education and student debt as an investment with an uncertain payoff. They conclude that the payoff to an undergraduate degree is worth the level of debt assumed for most students. In summary, the results concerning the impact of debt on educational outcomes are quite mixed and depend heavily on the sample of graduates examined and the outcomes investigated.

This article contributes to this literature by investigating the impact of state aid, admissions and financial aid policies, and student outcomes on variation in average student debt levels across institutions. The existing literature has primarily focused on the influence of loans on enrollment decisions or the impact of loans on educational outcomes at the individual level. Only one other study to my knowledge has examined variation across institutions in average student debt, but this study was not able to control for differences in state aid levels, institutional aid policies, and student outcomes and the role they play in determining average student indebtedness. This study fills that void.

Conceptual Framework

This analysis models nonprofit higher education institutions as attempting to maximize institutional prestige, subject to a nonprofit budget constraint. This framework of quality maximization is commonly used in analyzing nonprofit entities and educational institutions in particular (e.g., Breneman 1970; Brewer, Gates, and Goldman 2002; Clotfelter 1996; Garvin 1980; James 1990; Massy and Zemsky 1994; Melguizo and Strober 2007; Zemsky, Wegner, and Massy 2006). Clotfelter (1996) explicitly states that one of the distinctive features of nonprofit educational institutions (versus for-profit corporations and educational institutions) is their mission “to be the best.” The difficulty in attempting to achieve or even define this goal more succinctly is identifying what constitutes and determines institutional quality. Melguizo and Strober (2007) argue that institutional prestige is primarily derived through faculty members’ activities, such as publishing, but also note that institutions strive to maximize prestige more directly by developing innovative curricula, building technologically advanced facilities, and improving undergraduate education.

Within this framework of prestige maximization, institutions attempt to enroll the best entering class possible subject to resource constraints. One of the difficulties in what Duffy and Goldberg (1997) call “crafting a class” is identifying the dimensions that determine the quality of an entering class. Academic credentials, such as high school grades, class rank, and standardized test scores, are used to determine the quality of an institution’s entering class. As long as academically desirable students can be found along all points of the income distribution, institutions have an incentive to offer price discounts to low-income students of high ability in an attempt to encourage them to attend their institution (and maximize prestige). On the other hand, greater institutional resources and wealth can be used to enhance the prestige of a university by hiring top-notch faculty, building state-of-the-art facilities, and undertaking new programs. Colleges and universities weigh the marginal benefits of student quality with the marginal costs of additional scholarships needed to attract desirable students.

The budget constraint dictates that revenues from all sources (tuition, auxiliary enterprises, endowment income, gifts, etc.) must be greater than or equal to all expenditures, at least in the long run. Borrowing from Clotfelter (1996), a university’s budget constraint must equate total revenue and total expenses. To dramatically simplify the constraint by ignoring research funding and expenditures:

T + G = E + S

where T is tuition revenue, G is other revenue (such as grant income and endowment income), E is expenditures (instructional, administrative, etc.), and S is institutionally funded scholarships. In this framework, scholarships are assumed to be a percent of the tuition charged, such that S = jT, where j represents the discount rate, or percent of tuition transferred to students in the form of institutional scholarships. Substituting for S in equation (1) and rearranging terms leads to:

(1 - j) T = E - G

This equation indicates that net tuition revenue (net of scholarship aid) must be adequate to fund expenses in excess of revenue from other sources. Institutions with inadequate resources to cover expenses from nontuition revenue must generate sufficient net tuition revenue by either charging a high enough tuition, lowering the discount rate, or both.

Net tuition revenue can be written in terms of its funding sources by noting that:

L + So + FC = (1 - j) T

where L represents loans, So is outside scholarships, and FC is family contribution from income and savings. Writing equation (3) in terms of L leads to:

L = (1 - j) T - So - FC

This equation simply indicates that, ceteris paribus, loans are positively related to tuition and negatively related to outside scholarships and family resources.

Noting that the discount rate j = S/T, and solving equation (3) for S/T reveals that

j = (T - L - So - FC) / T

Thus, an institution can reduce its discount rate, and thus generate additional revenue, holding tuition constant, by increasing the level of loans it packages in its financial aid offers to students. Similarly, an institution could reduce its discount rate by seeking additional outside scholarship aid or admitting a more affluent student body.

This simplified university budget constraint predicts that an institution’s level of student indebtedness is positively related to its tuition and negatively related to internally funded and outside scholarship levels and an institution’s family resource profile. These attributes may be a function of institutional policies and practices. For example, meeting-full-need and being need-blind in admission may influence the family profile that an institution attracts and may have an impact on the level of contribution assumed to come from family resources.

Data

The data for this analysis come from the College Board’s 2011 Annual Survey of Colleges (ASC) merged with the 2011 Integrated Postsecondary Education Data System (IPEDS) database and state aid data from the 2010–2011 National Association of State Student Grant and Aid Programs’ (NASSGAP) Annual Survey Report on State-Sponsored Student Financial Aid. The ASC data consist of survey responses from 3,920 accredited undergraduate colleges and universities across the United States. To participate in the survey an institution must offer at least an associate degree and be accredited by a regional or national accrediting agency recognized by the U.S. Department of Education. According to the Digest of Education Statistics (National Center for Education Statistics 2012), there are 4,599 degree-granting institutions in the United States; thus, the Annual Survey of Colleges sample represents more than 85 percent of all degree-granting institutions nationwide.

To examine variation in average student debt across nonprofit four-year colleges and universities, I impose a number of sequential restrictions on the dataset. I eliminate for-profit institutions as they rarely reported average debt levels in this survey, and their institutional objectives may be quite different from those of the nonprofit higher education sector (n = 892 institutions). The sample also eliminates institutions whose highest degree awarded is an associate degree (n = 1,040), as comparing borrowing for two years of study to indebtedness for four years of education is inappropriate. I also eliminated observations where important variable values were not reported. Specifically, I omit observations where the institution did not report the percent of students with debt or the average debt (n = 1,000), percent of the class receiving financial aid or the reported percent was greater than 100 percent (n = 191), or percent of need met with financial aid and/or the cost of attendance (n = 50). These restrictions result in a final sample of 747 four-year, bachelor’s degree–granting, accredited institutions with valid data on the financial aid profile of their graduating classes.

The Annual Survey of Colleges questionnaire defines indebtedness as the “aggregate dollar amount borrowed through any loan program (federal, state, subsidized, unsubsidized, private, etc.; excluding parent loans) while the student was enrolled at an institution. Student loans co-signed by a parent are assumed to be the responsibility of the student and should be included.” Two measures of average student debt are used: average student debt for those with debt; and total average debt among all graduates, including those who did not borrow. These two measures of debt reflect the most aggregated student debt levels reported by institutions. The first is usually the number presented in the popular media. For example, a New York Times exposé on student borrowing (Martin and Lehren 2012) cites the 2011 Federal Reserve of New York report and asserts that, among borrowers, the average student debt in 2011 was $23,300. For the sample of institutions in this analysis, the average (across institutions) student debt among borrowers in 2011 was $24,646 (see Table 1 for summary measures) and the undergraduate enrollment weighted average student debt was $22,799. These results indicate that the restrictions placed on the sample outlined above do not skew the results toward higher or lower average levels of student debt. Total average debt, including nonborrowers, in the sample used in this article is $16,769 (across institutions), while the undergraduate enrollment weighted average is $13,772.

Table 1

Summary Measures

	Minimum	Maximum	Mean	SD
Percent of students with loans	1	100	65.89	16.83
Average debt among borrowers	$947	$56,000	$24,646	$7,677
Total average debt (incl. nonborrowers)	$119	$47,600	$16,769	$7,787
Public institution	0	1	0.36	0.48
State grant aid/FTE	$7	$1,895	$644	$402
Need-based state grant aid/FTE	$4	$976	$463	$291
Cost of attendance	$8,986	$58,334	$33,537	$12,616
Expenditures per student	$5,291	$270,834	$26,153	$22,728
Need-blind	0	1	0.88	0.33
Meet-full-need	0	1	0.08	0.26
Percentage of gap	0	98	24.86	16.74
No. of FTE undergraduates	129	46,994	5,644	6,805
SAT midpoint	760	1515	1108.79	135.66
Percent of students on financial aid	0	1	0.81	0.13
Percent high paying majors	0	1	.43	.16
Four-year graduation rate	0	92	44.13	22.07
Percent of students with Pell grants	6	100	30.59	13.88
Highest degree awarded:
Masters	0	1	0.36	0.48
Doctorate	0	1	0.47	0.50

The analysis also includes dummy variables indicating whether the institution admitted its 2010–2011 fall class in a need-blind manner (i.e., they did not consider an applicant’s financial aid profile in the admission decision) and met the full demonstrated need of all admitted students. Meeting-full-need means that all of the difference between the tuition, room, and board of an institution and what a family is expected to contribute for their child’s higher education expenses is met with grant aid, loans, or work-study. While a number of institutions report having admitted their class in a need-blind manner (88 percent), only 8 percent report meeting full demonstrated need of the entire entering class (see Table 1 for summary measures). Only thirty-eight institutions, or just over 5 percent of the institutions in this sample, claim to be both need-blind and meet-full-need. The degree to which an institution “gaps” its students, on average, is also used as a control variable. The expectation is that the more an institution gaps its students, the higher the level of student debt. As meeting-full-need can be achieved with varying degrees of loans versus grant aid, I also include dummy variables indicating whether an institution has a policy of no-loans for all students, such that all of a student’s demonstrated need is met with grant aid; and a dummy variable for whether an institution has a policy of limited loans, such that students below a certain income threshold have all of their need met with grant aid, as identified in The Project on Student Debt for 2009–2010 (2012). Just over 2 percent of the institutions in this sample have a no-loan policy and approximately 4 percent limit loans for low-income students.

Other institutional control variables are the percent of the entering class receiving financial aid, the percent of the entering class receiving Pell grants, and the median SAT score of the entering class. These variables are intended to capture the economic profile of the student body. The lower the percentages of the class receiving aid and Pell grants, the more affluent the class. Similarly, the higher the SAT score of the class, the more affluent the student body generally, given the high, consistent, positive correlation between family income and standardized test scores.

I use the total cost of attendance (i.e., tuition, fee, room, board, books and expenses) to measure the sticker price of the institution. For public universities I use an undergraduate enrollment weighted average of the in-state and out-of-state cost of attendance. I also include total expenditures per student to examine whether institutions with greater demands on their resources pass those expenses on to students in the form of higher debt packages, even within the framework of meeting-full-need.

I also include among the independent variables the four-year undergraduate graduation rate and the percent of all graduates who obtained degrees in majors with above-median earnings. I identified these majors from the Center on Education and the Work Force’s analysis of earnings by college major (Carnevale, Strohl, and Melton 2011). The majors with above-median earnings are engineering, mathematics and statistics, physical sciences, science technologies, social sciences, health professions, and business. Controls for public versus private institutions, highest degree awarded, and institutional size are also included among the regressors. These regressors are included among the independent variables to allow for variation across institutional type in the degree to which external pressures are brought to bear on institutions to control student debt levels. Public institutions and smaller, undergraduate-focused institutions may be under greater pressure from state legislatures to keep undergraduate student debt levels down rather than pursue institutional prestige.

I control for variation in state grant aid support by incorporating the amount of state-provided undergraduate grant dollars per FTE undergraduate matriculant in the state in which the college or university is located. Also included is the percent of undergraduate grant dollars per FTE awarded based on need (versus merit). This percentage varies from a low of 0.2 percent to a high of 100 percent. Institutions in states with more generous state aid grants are expected to have lower levels of student debt, as these grants can be passed along to students and used to lower borrowing. In two states with the same generosity of state grant aid, institutions in the state that allocate grants based on need are expected to have lower levels of average debt, as the state grant is going primarily to those who are most likely to borrow; non-need-based state grants go to students across the income distribution.

All average debt levels, cost of attendance, and expenditure per student values are in natural log form, so that the coefficients on the natural log of costs, expenditures, and state aid per student indicate the elasticity of student debt with respect to these variables, while the coefficients on the other variables approximate the percentage change in student debt for a unit change in the independent variables. All regressions are estimated using weighted least squares, where the weight is FTE undergraduate to account for the varying institution sizes used in calculating the average debt levels.

Results

The regression analyses investigate the influence of price, state aid, admissions practices, financial aid policies, and student outcomes on the average student debt of the graduating classes across nonprofit, four-year postsecondary institutions in the United States.

Column 1 in Table 2 presents the results of the regression of the natural log of average student debt among borrowers on the exogenous variables outlined above. Before turning to the variables of primary concern, note that there is no difference in average indebtedness among borrowers between public and private institutions. Institutional size (number of FTE undergraduates) and type (baccalaureate, masters, doctoral) are also not significant in determining average student indebtedness, conditional on the other variables. The quality of the institution as measured by the SAT midpoint is significant in determining the average student debt among borrowers. Institutions with higher average SAT scores have lower levels of student borrowing. This finding is likely attributable to the strong correlation between student SAT scores and family income.

Table 2

Regression Results: Dependent Variable Is Natural Log of Average Debt

	Among Borrowers	All
	Borrowers	Students
Constant	7.215***	6.342***
	(0.771)	(1.071)
Natural log of cost of attendance	.229 ***	.292***
	(0.078)	(0.108)
Natural log of expenditures per student	.126***	.115**
	(0.037)	(.052)
Natural log of state grant aid/FTE	−.046***	−.085***
	(0.015)	(0.021)
Percent state aid that is need-based	−.006	.051
	(.042)	(0.059)
Need-blind	.275***	.357***
	(0.041)	(0.057)
Meet-full-need	−.184**	−.292***
	(.073)	(0.102)
Percent gap	−.00002	−.001
	(.001)	(0.001)
Percent of students with aid	.001	.008***
	(.001)	(0.002)
No loan policy for need-based aid	−.655***	−.902***
	(0.114)	(0.159)
Limited loan policy for need-based aid	−.138***	−.243***
	(0.043)	(0.059)
Four-year graduation rate	.004***	.008***
	(0.001)	(0.002)
Pct. of grads in high-paying majors	.298***	.429***
	(0.110)	(0.153)
Pct. of students receiving Pell grants	−0.003**	−.001
	(0.002)	(0.002)
No. of FTE undergraduates (in 1,000s)	−.001	−.004
	(0.002)	(0.002)
SAT midpoint (in 100s)	−.095***	−.165***
	(0.023)	(0.031)
Publicly controlled	−.044	−.068
	(0.058)	(.081)
Masters awarding university	.085	.065
	(0.065)	(0.091)
Doctoral awarding university	.051	.035
	(0.066)	(0.091)
R-squared	.314	.398
No. of observations	747

NOTE: Unstandardized coefficients; standard errors are in parenthesis.

***

significance at the 1% level; **significance at the 5% level; *significance at the 10% level.

Turning to the variables and policies of primary interest, it is not surprising that the cost of attendance is significant in determining student debt. However, cost is inelastic in determining student debt. The elasticity of student debt with respect to tuition is only .229. This finding implies that a 10 percent increase in the cost of attendance results on average in a 2.3 percent increase in average student debt. With increases in institutional expenditures per student come increases in average student borrowing. It appears that institutions with higher levels of spending per student pass at least some of these higher expenses on to their students in the form of increased levels of student debt, conditional on the other variables. Once again this effect, although statistically significant, is inelastic in magnitude.

State grant aid is negatively related to student debt among borrowers, although once again the magnitude of the effect is inelastic. A 10 percent increase in state aid per undergraduate is associated with approximately a .5 percent lower student debt on average. The proportion of undergraduate state-administered aid that is disbursed based on need is unrelated to student debt levels among both borrowers and all students.

Admissions and financial aid policies are related to student borrowing. Institutions that report being need-blind in admissions have average student indebtedness for borrowers that is approximately 32 percent (exp(.275)-1) higher than comparable institutions that are not need-blind in admissions. Evaluated at $24,000 in student debt, this coefficient translates to approximately $7,700 in additional borrowing at need-blind institutions compared to those that are not need-blind in admitting their freshmen class, conditional on the other variables. Although usually considered beneficial in providing greater opportunities for students from lower-income families, need-blind admission appears to have the unintended consequence of leading to higher average debt levels, as more low-income students constitute the subsequent graduating class.

Institutions that meet the full demonstrated need of their students have average student debt that is approximately 17 percent (exp(–.184)-1) lower than institutions that do not meet-full-need. This finding is not surprising, as students at institutions that gap may have to turn to additional loans to meet their education expenses. Surprisingly, the percent of gapping by an institution is unrelated to student debt at graduation. Perhaps families turn to other forms of funding such as home equity or PLUS loans in the presence of substantial gapping. It is also surprising that the percent of students receiving financial aid is unrelated to average debt among student borrowers, but the percent of students receiving Pell grants is negatively related to the level of debt among borrowers. The findings suggest that, conditional on the other regressors, higher shares of Pell grant recipients are associated with lower debt needed to meet educational expenses. Pell grants may be replacing loans for low-income students, or institutions that attract more Pell recipients may be limiting loan levels.

The results imply that need-blind institutions have higher levels of average student debt, while institutions that meet-full-need and have higher levels of state aid have lower levels of student debt. Meeting-full-need may be manifest differently across institutions. For example, a student with demonstrated need of $20,000 may have all of that need met with grants at one institution, but met with loans and grants at another. Clearly the borrowing implications of these institutions that meet-full-need are quite different. To account for these differences I include a dummy variable for institutions that have a no-loan policy for all students on need-based aid, and a separate dummy variable for institutions that have a limited-loans policy whereby students below a particular income threshold have all need met with grant aid. Even with these policies, students, of course, are free to borrow. Nonetheless, the results show that no-loan and limited-loan policies are related to student borrowing. Compared with comparable institutions, average student debt is 48 percent lower at institutions with no-loan policies and 13 percent lower at institutions with a limited-loan policy.

Pricing and financial aid policies are not the sole institutional criteria related to average institutional debt levels. As expected, the higher the likelihood of graduation and the better the employment prospects of an institution’s graduates, the higher the level of average debt.

In short, the cost of attendance appears to have a statistically significant but inelastic effect on student debt, while institutional policies of being need-blind, meeting-full-need, and restricting or even eliminating loans are all associated with the average student debt levels among borrowers. Increases in state aid are associated with lower levels of student debt, but the magnitude of the relationship is quite small. The expected academic and labor market success of an institution’s graduates also are related to student indebtedness.

Total average student debt among all graduates from an institution, including nonborrowers, is perhaps a more meaningful and comprehensive measure of the debt burden placed on an institution’s graduates. For example, an institution where only 20 percent of students borrow but that has average debt among borrowers of $30,000 is arguably placing less of a loan burden on its graduates than an institution where 80 percent of its graduates have an average debt of $25,000. Column 2 of Table 2 regresses the natural log of the total average student debt (including nonborrowers) against the independent variables. Although many of the results are qualitatively similar to the previous results, the magnitudes of the relationships are often even larger than for the analysis of debt among borrowers.

Once again, the cost of attendance and expenditures per student are positively related to student debt, but the relationships are inelastic. State grant aid is inversely related to average student borrowing, and the magnitude is twice as large, although still inelastic. Being need-blind and meeting-full-need are also significantly associated with total average debt; the magnitudes of the relationships are noticeably larger than for the analysis of borrowers. Similarly, the percent of students with aid has a positive and even larger relationship to total debt than average debt among borrowers. Institutions with no-loan and limited-loan policies have lower levels of total student debt as well. Graduation rates and the mix of majors are once again related to variation across institutions in total average student debt, with amplified effect. In short, all of the variables that were significantly related to average student debt among borrowers are also significantly related to total average student debt among all students, with the exception of the percent of students receiving Pell grants, which is no longer significant in determining debt. The primary difference in the results is the size of the coefficients. In particular, the coefficients for the financial aid and academic outcome variables are substantially larger in magnitude.

The analyses implicitly assume that the impact of costs and aid policies on debt levels is the same across public and private higher education sectors. As public and private institutions may face different pressures from the general public and state legislatures to control student debt, and as state grant aid policies may have differential impacts across sectors, separate analyses for public and private institutions are warranted. Chow-tests reject the null hypothesis of equality of the coefficients across the public and private sectors (at the 1 percent level) for the regression using the natural logs of total average student debt (F-stat = 2.94).

Table 3 presents the results of the regressions of the natural log of total average student debt for all students on the independent variables for public institutions (column 1) and private institutions (column 2). The results in some instances vary dramatically between the two sectors.

Table 3

Regression Results: Dependent Variable Is Natural Log of Average Debt for All Students

	Public	Private
Constant	8.802***	2.218***
	(2.121)	(0.973)
Natural log of cost of attendance	−.016	.861***
	(0.222)	(0.102)
Natural log of expenditures per student	.143	−.069
	(.090)	(.053)
Natural log of state grant aid/FTE	−.108***	−.035
	(0.036)	(0.024)
Percent state aid that is need-based	.055	−.009
	(0.104)	(0.057)
Need-blind	.449***	.057
	(0.105)	(0.052)
Meet-full-need	−.673	−.196***
	(0.415)	(0.070)
Percent gap	−.001	.0001
	(0.002)	(0.001)
Percent of students with aid	.007**	.009***
	(0.003)	(0.002)
No loan policy for need-based aid	---------	−.613***
		(0.096)
Limited loan policy for need-based aid	−.280***	0.003
	(0.098)	(0.088)
Four-year graduation rate	.010***	.001
	(0.003)	(0.002)
Pct. of grads in high-paying majors	.602	.257**
	(0.354)	(0.106)
Pct. of students receiving Pell grants	−.001	.002
	(0.004)	(0.002)
No. of FTE undergraduates (in 1,000s)	−.004	−.003
	(0.004)	(0.004)
SAT midpoint (in 100s)	−.157***	−.150***
	(0.059)	(0.029)
Masters awarding university	.236	.0001
	(0.391)	(0.054)
Doctoral awarding university	.184	.061
	(0.387)	(0.055)
R-squared	.276	.653
No. of observations	271	476

NOTE: Unstandardized coefficients; standard errors are in parenthesis.

***

significance at the 1% level; **significance at the 5% level; *significance at the 10% level.

At public universities, the cost of attendance is not related to total average student debt, conditional on other variables. This result implies that a focus on sticker price in determining debt levels, at least among graduates of public universities, may be misplaced. On the other hand, state grant aid is negatively related to total average debt. A 10 percent increase in average state grant aid per student is associated with just over a 1 percent reduction in average student debt at public universities, conditional on the other variables. The mix of state grant between need-based and merit-based aid does not appear to be related to total average student debt at public universities.

Being need-blind is associated with higher levels of average student debt. Public universities that are need-blind in admissions have average student debt that is approximately 57 percent higher (exp(.449)-1) on average than comparable public institutions that are need-aware in admissions. On the other hand, meeting-full-need is related to a 49 percent drop (exp(–.673)-1) in total average student debt at public universities. Although the magnitude of the relationship is quite large, it is not statistically significant as there are few public universities with this policy.

The percent of students on financial aid is positively associated with the average level of total student indebtedness. While there are no public universities with no-loan policies, the public universities with limited loan policies have average total student debt levels 24 percent lower on average than public universities that did not eliminate loans for low-income students. The percent of students on financial aid is unrelated to total average debt in this sector.

Higher graduation rates are also associated with higher total average debt levels. Similarly, a higher percent of degrees awarded in high paying majors is also associated with higher average debt levels, but the coefficient is not significantly different from zero at conventional levels at public institutions. The only other regressor that is significantly correlated with total average student debt levels at public institutions is a school’s average SAT score. A 100 point increase in average SAT score is correlated with a 15.7 percent decline in total average student debt.

Institutional size, as measured by the number of FTE undergraduate students, is not related to average student debt. There are no significant differences in total average student debt levels across types of institutions (doctoral, masters, and baccalaureate) within the public higher education sector.

In summary, cost of attendance is not related to total average student debt at public universities conditional on the other regressors. On the other hand, state grant aid is associated with a significant reduction in total average student debt levels at public universities. Public institutions with limited-loan policies also have lower average student debt, while universities that are need-blind in admissions have higher levels of average student debt. More selective public universities as measured by average SAT scores have significantly lower levels of average student debt.

While qualitatively similar for public and private institutions, the results have notable differences. Particularly, the cost of attendance has a positive and statistically significant relationship with total average student debt at private colleges and universities. In this case, a 10 percent increase in the cost of attendance is associated with an 8 percent increase in total average student debt. This finding suggests that an institution with total average student debt of $15,000 upon graduation would see average debt increase by approximately $1,300 for a 10 percent increase in the cost of attendance. This result is not significantly different from unit elasticity.

State grant aid, however, is not related to total average student debt at private institutions. The mix of need-based versus merit-based state grant is also not related to variation in total average student debt levels at private institutions.

Being need-blind in admissions is not related to total average student borrowing, but meeting-full-need is negatively related to total average student debt at private institutions. Institutions that meet full need have total average student debt almost 18 percent lower on average than institutions that gap their students. As was the case with public institutions, the higher the percent of students receiving financial aid, the higher the level of total average student debt at private institutions as well.

Adhering to a no-loan policy is associated with lower levels of total average student debt in the private higher education sector. Private institutions with a no-loan policy have total average student debt approximately 46 percent lower than institutions that do not eliminate loans for their students. A limited loan policy is not related to total average student debt levels, perhaps because this practice is rare even among private institutions.

Although the coefficient on the four-year graduation rate is positive at private colleges and universities, it is not significantly different from zero at conventional levels. The percent of degrees awarded in high-paying majors, however, is associated with higher levels of total average student debt. A 10 percentage point increase in the number of degrees awarded in high-paying fields is associated with approximately a 3 percent increase in total average student debt. Once again, average SAT scores are negatively related with lower levels of total average student debt.

At private colleges and universities, the cost of attendance is related to total average student debt levels, while state grant aid is not. These are the primary differences between public and private institutions. State grant aid appears to have a significant effect in ameliorating total average student debt levels at public institutions, while its impact is much more modest at private institutions.

Conclusion

Recent media coverage of growing student debt and the subsequent policy proposals to address it have assumed that tuition and university sticker prices are the primary if not sole force driving the rise in student indebtedness. This assumption ignores the substantial impact that state and university admissions and financial aid policies and academic outcomes play in determining variation in average student debt levels across institutions. This article finds that, although the cost of attendance is significantly related to student debt levels at private institutions, state aid, institutional admissions and financial aid policies, graduation rates, and the mix of majors across students are also important sources of institutional variation in average student debt levels. Specifically, whether an institution is need-blind in admitting its students or whether it meets the full demonstrated need of all students can increase average student debt upon graduation by as much as 30 percent. The impact of these policies varies substantially between public and private sectors. Being need-blind and the level of state grant aid are related to student borrowing at public institutions, while meeting-full-need is negatively related to student debt at private institutions. No-loan and limited-loan policies are also important in determining student debt levels even conditional on broader financial aid policies.

Public universities that adhere to a need-blind admissions policy have higher average student debt upon graduation, ceteris paribus. A need-blind admissions policy is usually viewed as a positive and beneficial practice that increases the probability of enrolling more low-income students. Although this policy may be effective in increasing low-income enrollment, it also is associated with higher levels of average debt upon graduation, as the mix of students at a particular institution becomes less affluent, unless coupled with financial aid policies that help to reduce average student debt, such as meeting-full-need and limiting loan levels. Penalizing institutions for having higher levels of student debt may lead them to adopt admissions and financial aid policies that are less conducive to enrolling low-income students. Institutions, both public and private, may become more selective in admissions with a heightened focus on SAT scores, which are highly positively correlated with family income, and thus negatively correlated with student debt. Similarly, institutions may no longer practice need-blind admissions, but rather turn to need-aware admissions practices to manage the income profile and subsequent borrowing levels of their student bodies. Although leading to lower levels of student debt, these practices may not be beneficial for social welfare.

State governments also may be partly culpable in rising average student debt levels. The Joint Economic Committee of the United States Congress’ (2013) report on the increasing levels of student debt hypothesizes that the decline in per-student state-based subsidies and grants may have forced many families to rely more heavily on federal loans. The results of this analysis support that assertion, as decreases in state-sponsored grants are associated with higher average student debt levels, at least among public university graduates.

The results of this article indicate that policy advocates interested in rising college costs and escalating student debt need to fully appreciate the interaction of tuition, state aid, and admissions and financial aid policies in determining variation in average student debt levels across institutions. Recommendations and policy proposals designed to limit student debt should focus on not only tuition levels and tuition inflation, but also state and institutional admissions and financial aid practices that affect net price and encourage students to pursue fields with an adequate return on investment despite the level of debt. Matching institutional grant aid with state grant aid may offer an effective policy option to hold student debt levels down. President Obama’s recent proposal (White House 2013) challenging states to fund public colleges based on performance and affordability metrics, if implemented properly, may also result in lower levels of average student debt. Policies that solely focus on tuition levels and tuition growth are overly simplistic and ignore other important factors in determining student debt.

Footnotes

Notes

James Monks is an associate professor of economics at the University of Richmond. He is also an affiliated faculty member with the Cornell Higher Education Research Institute. His research focuses on the economics of higher education and includes publications in Economic Inquiry, Industrial Relations, and Economics of Education Review.

References

Avery

Christopher

Turner

Sarah

. 2012. Student loans: Do college students borrow too much—or not enough? Journal of Economic Perspectives 26 (1): 165–92.

Breneman

David

. 1970. The Ph.D. production function: The case of Berkeley. Ford Foundation Program for Research in University Administration Paper. Berkeley, CA: Ford Foundation.

Brewer

Dominic J.

Gates

Susan M.

Goldman

Charles A.

2002. In pursuit of prestige: Strategy and competition in U.S. higher education. New Brunswick, NJ: Transaction Publishers.

Buss

Christian

Parker

Jeffrey

Rivenburg

Jon

. 2004. Cost, quality, and enrollment demand at liberal arts colleges. Economics of Education Review 23 (1): 57–65.

Carnevale

Anthony

Strohl

Jeff

Melton

Michelle

. 2011. What it’s worth: The economic value of college majors. Washington, DC: Georgetown Center on Education and the Workforce.

Cauchon

Dennis

. 18 October 2011. Student loan debt surpasses $1 trillion. USA Today. Available from http://www.usatoday.com.

Clotfelter

Charles

. 1996. Buying the best: Cost escalation in elite higher education. Princeton, NJ: Princeton University Press.

College Board. 2011. Trends in student aid, 2011. New York, NY: The College Board. Available from http://trends.collegeboard.org.

College Board. 2012. Trends in college pricing, 2012. New York, NY: The College Board. Available from http://trends.collegeboard.org.

10.

Craven

Julia

. 5 July 2013. Universities attempting to lower student loan debt. USA Today. Available from http://www.usatoday.com.

11.

Delaney

Jennifer A.

2014. The role of state policy in promoting college affordability. The ANNALS of the American Academy of Political and Social Science, (this volume).

12.

Duffy

Elizabeth A.

Goldberg

Idana

. 1997. Crafting a class: College admissions and financial aid, 1955–1994. Princeton, NJ: Princeton University Press.

13.

Field

Erica

. 2009. Educational debt burden and career choice: Evidence from a financial aid experiment at NYU Law School. American Economic Journal: Applied Economics 1 (1): 1–21.

14.

Garvin

David A.

1980. The economics of university behavior. New York, NY: Academic Press.

15.

James

Estelle

. 1990. Decision processes and priorities in higher education. In The economics of American universities, eds. Hoenack

Stephen A.

Collins

Eileen L.

, 77–106. Buffalo, NY: State University of New York Press.

16.

Joint Economic Committee of the United States Congress. 2013. The causes and consequences of increasing student debt. Prepared by the Joint Economic Committee Democratic Staff, G-01 Dirksen Senate Office Building, Washington, DC.

17.

Linsenmeier

David M.

Rosen

Harvey S.

Rouse

Cecilia E.

2006. Financial aid packages and college enrollment decisions: An econometric case study. Review of Economics and Statistics 88 (1): 126–45.

18.

Luong

May

. 2010. The financial impact of student loans. Perspectives on Labour and Income 22 (1): 29–42.

19.

Macy

Ann

Terry

Neil

. 2007. The determinants of student college debt. Southwestern Economic Review 34 (1): 15–25.

20.

Martin

Andrew

Lehren

Andrew W.

13 May 2012. A generation hobbled by college debt. New York Times.

21.

Massy

William F.

Zemsky

Robert

. 1994. Faculty discretionary time: Departments and the academic ratchet. Journal of Higher Education 65 (1): 1–22.

22.

Melguizo

Tatiana

Strober

Myra H.

2007. Faculty salaries and the maximization of prestige. Research in Higher Education 48 (6): 633–68.

23.

Minicozzi

Alexandra

. 2005. The short term effect of educational debt on job decisions. Economics of Education Review 24 (4): 417–30.

24.

Monks

James

. 2001. Loan burdens and educational outcomes. Economics of Education Review 20 (6): 545–50.

25.

National Association of State Student Grant and Aid Programs. 2002. 32^nd annual survey report 2000-2001 academic year: State-funded scholarship/grant programs for students to attend postsecondary education institutions. Washington, DC: National Association of State Student Grant and Aid Programs.

26.

National Association of State Student Grant and Aid Programs. 2012. 42^nd annual survey report on state-sponsored student financial Aid. Washington, DC: National Association of State Student Grant and Aid Programs.

27.

National Center for Education Statistics, Department of Education. 2012. Digest of education statistics, 2011. Washington, DC: National Center for Education Statistics.

28.

The Project on Student Debt. 2012. Summary of pledges: Eligibility guidelines and basic provisions (2009–10). Available from http://projectonstudentdebt.org.

29.

Reynolds

Glenn Harlan

. 28 June 2013. What’s really “immoral” about student loans. Wall Street Journal. Available from http://online.wsj.com/news/articles/.

30.

Rothstein

Jesse

Rouse

Cecilia E.

2011. Constrained after college: Student loans and early-career occupational choices. Journal of Public Economics 95:149–63.

31.

Tumen

Sarah

Shulruf

Boaz

. 2008. The effect of student loan schemes on students returning to study. Journal of Higher Education Policy and Management 30 (4): 401–14.

32.

Waddell

Glen R.

Singell

Larry D.

2011. Do no-loan policies change the matriculation patterns of low-income students? Economics of Education Review 30 (2): 203–14.

33.

Wenzl

Roy

. 10 August 2013. Crippling student debt affects economy and that’s bad news for everyone, expert at KU says. The Wichita Eagle. Available from http://www.kansas.com.

34.

The White House. 22 August 2013. Fact sheet on the President’s plan to make college more affordable: A better bargain for the middle class. Washington, DC: Office of the Press Secretary of the White House. Available from http://www.whitehouse.gov.

35.

Zemsky

Robert

Wegner

Gregory

Massy

William F.

2006. Remaking the American university: Market-smart and mission-centered. New Brunswick, NJ: Rutgers University Press.