Morrie Swerlick Student Debt Policy Memo 2/23/2012 Student debt from higher education attendance is an increasingly troubling problem in the United States. Due to rising costs and shrinking state expenditures, students and their families are being forced to shoulder increasing burdens when it comes to higher education. This became increasingly evident when the total level of Student debt surpassed $1 trillion, an amount that exceeds even that of credit card debt in the United States (Pilon 2010). Combined with a weak economy and job market, student debt levels have become an issue of national conversation. Despite the issues pertaining to college costs, the job prospects of college graduates remain stronger than those of individuals with only a high school education (Project on Student Debt 2010). With this in mind, putting the pieces together in the student debt puzzle is an important part of teaching necessary skills to gain employment and help the economy. According to a report published by the Project on Student Debt in 2010, the average amount of debt for graduating student in the class of 2010 was $25,250. The analyses discussed below use 2004-2005 data from the Student Debt Project, which contains institutional level data of 1093 colleges and universities (both 4 year-public and 4 year- private). While somewhat outdated, this data is still useful in exploring the factors which effect student debt. On average, 66% of a school s 2004-2005 graduating class left with some level of debt. This number appears to be holding steady based on 2008 data and projections for 2010 (Project on Student Debt 2010). The average total debt at graduation across these schools in 2004-2005 was $18,174.71 with a standard deviation of $5,401.31. As evident by the large increase from 2005 to 2010, student debt levels are growing rapidly. The data set also includes data on various factors which potentially have an impact on debt levels at college and universities. One factor is the various types of financial aid used to finance an
education. The survey collected information on the percentage of students who applied for aid in general, as well as the percentage of students who received Pell grants, subsidized loans, and unsubsidized loans. Across all schools, the average percentage of dependent students applying for aid was 69.16%. Broken down into types of aid, 27.5% received Pell grants, 45.2% received subsidized loans, and 28.55% received unsubsidized loans. As mentioned above, the data set includes both public and private schools which vary in their selectivity (Very Selective, Moderately Selective, Minimally Selective, and Open Admission). Debt levels differ across both institution type and selectivity. In line with conventional wisdom, Private institutions graduated a significantly higher percentage of students with debt and had the highest average student debt loads when compared to Public institutions. This is not surprising, considering that tuition and fees at Private institutions are higher than at Public schools. However, counter to conventional wisdom, highly selective institutions did not show significantly higher levels of average graduate debt when compared to all other levels of selectivity. In fact, in the public sector, there was no significant difference between the average amounts of debt or percent of students graduating with debt across institution selectivity. Within private institutions, Moderately Selective schools had higher average total debt levels and percent of students who graduated with debt than very selective institutions. In order to try and understand the effects that variables such as sector, selectivity, and type of aid had on both the average amount of debt for graduating students and on the percentage of student who graduate with debt, linear regressions were run. The initial model included institution sector, percent of students receiving Pell grants, percent of students receiving subsidized loans, and controlled for moderately selective schools. Moderately selective schools were chosen instead of Very Selective because previous analysis showed potential evidence that moderately selective school had higher levels
of debt and percentage of students graduating with debt. Percent of students who graduate with debt was used as the outcome variable. Model Model Summary R R Square Adjusted R Square Std. Error of the Estimate 1.617 a.381.377 13.251 a. Predictors: (Constant), Sector,, Minimally Selective, Open Admission, Very Selective, month enrolled Model 1 Standardized Coefficients B Std. Error Beta t Sig. (Constant) 41.272 1.380 29.916.000 Very Selective -4.796 1.060 -.120-4.524.000 Minimally Selective 1.902 1.229.039 1.547.122 Open Admission.749 1.715.011.437.663 Unstandardized Coefficients.569.036.548 15.773.000 -.139.043 -.109-3.233.001 Sector 5.910.974.168 6.066.000 This initial model accounts for 38% of the variance in the percent of students graduating from college with debt. The variables for Very Selective institutions, Private Schools, subsidized loans, and Pell grants were found to be significant predictors. These findings somewhat line up with the results of previous analysis discussed earlier, with Private institutions showing significantly higher percentage of students graduating with debt (5.9% increase associated with a school being private).the only selectivity level which was found to be significant was Very Selective. Compared to Moderately selective institutions, Very Selective schools graduate 4.8% less students with debt. Both financial aid variables were significant in the model. For each 1% increase in the percentage of 12 month enrollees at a school receiving Pell Grant support the percentage of students graduating with debt decreased.14%. In regards to Subsidized loans, each 1% increase in the percent of 12 month enrollees receiving these loans was
associated with a.57% increase in the percentage of students graduating with debt. The direction of these findings fit with the characteristics of the types of aid. Pell grants, do not require payment, and would logically reduce the percent of students graduating with debt. Subsidized loans are a source of aid which requires eventual repayment, and thus the expected effect would be opposite. However, as discussed above the percentage of students graduating with debt differs between public and private institutions, with private schools having a significantly higher percentage of graduate leave with debt. With this in mind, it makes sense to also look at the data for public and private schools separately. Model Summary Sector Model Adjusted R Std. Error of the R R Square Square Estimate Public 1.567 a.322.313 13.721 Private 1.566 a.320.315 12.566 a. Predictors: (Constant),, Minimally Selective, Open Admission, Very Selective, m Sector Public 1 Model Standardized Coefficients B Std. Error Beta t Sig. (Constant) 33.328 2.361 14.114.000 Very Selective.344 1.962.008.176.861 Minimally Selective 5.135 1.944.118 2.642.009 Open Admission 8.455 2.719.143 3.109.002 Unstandardized Coefficients.935.075.713 12.406.000 -.415.084 -.299-4.940.000 Private 1 (Constant) 52.253 1.811 28.848.000 Very Selective -7.764 1.224 -.223-6.342.000 Minimally Selective.703 1.558.015.451.652 Open Admission -2.876 2.179 -.044-1.320.187.439.040.469 11.007.000 -.048.049 -.043 -.981.327 a. Dependent Variable: PercentwDebt
While the separate models for public and private schools explain similar amounts of variance (32% for both), the factors which are significant differ. In Public institutions, both being classified as Open Admission and Minimally selective are associated with increases in the percent of students who graduate with debt when compared to moderately selective schools (8.4% and 5.1% respectively). Within the private sector, Very selective is the only significant selectivity factor, with a 7.8% decrease in the percentage of students graduating with debt compared to moderately selective schools. Additionally, while a 1% increase in the percent of students receiving Pell grants at public schools was associated with a.4% decrease in the percent of student who graduate with debt, the variable was not significant in Private schools. Subsidized loans were significant within both sectors, with a 1% increase in students receiving these loans associated with a.9% increase of students graduating with debt at Public schools and a.4% increase at Private schools. Model Summary Sector Model Adjusted R Std. Error of the R R Square Square Estimate Public 1.443 a.197.186 3.83688421 Private 1.354 a.125.119 5.18394484 a. Predictors: (Constant),, Minimally Selective, Open Admission, Very Selective, Subsidized loans: % of 1
Sector Public 1 Model Standardized Coefficients B Std. Error Beta t Sig. (Constant) 12.758.660 19.320.000 Very Selective -.817.549 -.072-1.489.137 Minimally Selective.239.543.021.439.661 Open Admission -.004.760.000 -.005.996 Coefficients a Unstandardized Coefficients.194.021.577 9.224.000 -.146.023 -.411-6.225.000 Private 1 (Constant) 17.841.747 23.874.000 Very Selective -1.270.505 -.100-2.514.012 Minimally Selective -.747.643 -.042-1.163.245 Open Admission -.620.899 -.026 -.690.490.133.016.391 8.092.000 -.166.020 -.413-8.262.000 When the outcome variable of the model is changed from the percentage of students graduating with debt to the average debt of graduates, the model explains 20% of the variance in public schools and 13% at private schools. Percent of students receiving Pell grants and subsidized loans were both significant factor. For each percent increase in the percent of students receiving Pell grants, the model projects a $146 decrease in average student debt. For Subsidized loans, each percent increase is associated with a $194 increase in average debt. Selectivity were found to be non-significant when only looking at public schools. In private schools, when compared to moderately selective institutions, very selective schools show less graduating student debt on average ($1,270). Additionally, the portion of student receiving both types of aid looked at were found to be significant predictors in the model. For each percent increase in the students receiving Pell grants and subsidized loans, the model shows an expected decrease of $166 (Pell grants) and an increase of $133 (Subsidized Loans) in student debt levels.
The results outlined above bring up some areas of consideration for policy makers. First, with public schools being cheaper in mind, it may be helpful to compile data on both input and output information across all institutions. Giving perspective students a better idea of what they are getting for their tuition dollars may allow them to find better value with public schools that have good output indicators for a lesser price than their private counterparts. Having this information may make students more comfortable attend schools which may be perceived as less prestigious or valuable based on conventional wisdom. Secondly, within some circles there have been calls for well endowed, highly selective institutions to spend down endowments (Lederman 2008).This suggestion does not align with the findings from previously mentioned analysis that suggest moderately selective institutions may have higher average debt loads and percentage of students graduating with debt. This could possibly signal that well-endowed, highly selective institutions are using endowments as a way of offsetting costs for some students (Lederman 2008). With this in mind, calls to spend down endowments may only represent a short term solution to controlling price, with long term negative effects. Further analysis of the issue should be done before taking any definitive steps in this realm. Lastly, the data collected suggests that providing grants over loans is preferable when trying to address issues of student debt. Unfortunately, a weak economy and shrinking appropriations for higher education make offering large amounts of grants to offset price increases difficult. Policy makers should consider exploring systems which incentivize student receiving grant money to attend public colleges and universities, where in many cases grant money could cover a larger portion of the cost associated with an education.
Chart and Tables Descriptive Statistics of Variables Used Average debt of graduates (2004-05) N Minimum Maximum Mean Std. Deviation 1093 1623 49337 18174.71 5410.309 1093 4 90 27.52 13.128 Percent of graduates with debt (2004-05) 1093.01 1.00.6620.16792 1093 1 98 45.20 16.168 Tuition and fees (2004-05) 1093 516 34030 13881.57 8479.716 Unsubsidized loans: % of 12-1093 3 81 28.55 12.580 Valid N (listwise) 1093 T-Test between Public and Private Institutions Tuition and fees Equal (2004-05) variances Equal variances not Percent of graduates with debt (2004-05) Equal variances Equal variances not Average debt of Equal graduates (2004-variances 05) Equal variances not Sig. (2- F Sig. t df tailed) 387.832.000-43.669 1091.000-56.714 863.451.000 3.356.067-12.697 1091.000-12.381 738.174.000 27.244.000-11.551 1091.000-12.448 973.814.000
Works Cited The Project on Student Debt. (2011). Student Debt and the Class of 2010. Retrieved from http://projectonstudentdebt.org/files/pub/classof2010.pdf. Pilon, Mary. (2010, August 9). Student Loan Debt Surpasses Credit Cards. The Wall Street Journal. Retrieved from http://blogs.wsj.com/economics/2010/08/09/student-loan-debt-surpasses-creditcards/ Lederman, Doug. (2008). Snapshots of Endowment spending. Inside Higher Ed. Retrieved from http://www.insidehighered.com/news/2008/03/07/endow