A Look at Unmet Financial Need by Family Income

One of the perks of my job is that I get to talk with journalists around the country on a regular basis—it gives me the chance to keep up on what are the hot topics in the broader community as well as build connections with some wonderful people. I recently chatted with Jeff Selingo of The Washington Post for his latest column on whether college is affordable for middle-class families. My quote in the piece was, “They are getting squeezed on both ends because they barely miss Pell Grants and they are not the types of students getting grants from colleges themselves.”

Because I’m a data person at heart, I wanted to provide some supporting evidence for my claim. I used the most recent wave of the Beginning Postsecondary Students Longitudinal Study—a nationally representative study of first-time college students in the 2011-12 academic year—to look at financial need among new students at four-year colleges by family income quintile (for dependent students, who are mainly traditional-aged). The key column in the table below is unmet financial need, which is how much money students and their families have to come up with to cover the cost of attendance after grant aid and the expected family contribution (EFC)—a rough estimate of how much the government thinks families can contribute.

Quintile Unmet need EFC Total grants Parent income
Bottom $10,000 $0 $9,318 $13,150
Second $10,637 $557 $8,550 $34,238
Middle $9,912 $5,440 $5,550 $61,388
Fourth $4,820 $14,537 $2,750 $95,763
Top $0 $31,663 $2,000 $161,361


Source: NPSAS 2011-12.

Note: Values presented are medians and are only for dependent students attending four-year colleges.

The key point here is that families in the middle income quintile have to come with roughly the same amount of additional money beyond the EFC to pay for a year of college as families in the bottom two quintiles. Grant aid drops off substantially after the second quintile (where Pell eligibility starts to phase out), so middle-income families certainly do have reasons to be concerned about college affordability. Federal loans and PLUS or private loans can help to bridge the gap for students, but these figures do illustrate why student debt burdens (although relatively modest from a lifetime perspective) are a mounting concern for a larger percentage of undergraduate students.

Posted in Uncategorized | Tagged , , | Leave a comment

Which Factors Affect Student Loan Repayment and Default Rates?

As student loan debt has surpassed $1.25 trillion, policymakers and members of the public are increasingly concerned about whether students are able to manage rising (but often still modest) loan burdens. The federal government has relied on a measure called cohort default rates—the percentage of former borrowers who defaulted on their loans within a few years of entering repayment—to deny federal financial aid access to colleges with a high percentage of struggling students. Yet default rates can be easily manipulated using strategies such as deferment and forbearance (which often don’t help students in the long run), meaning that default rates are a very weak measure of students’ post-college outcomes.

The 2015 release of the College Scorecard dataset included a new measure—student loan repayment rates, defined as the percentage of borrowers repaying any principal within a certain period of entering repayment. This gets at whether students are paying down their loans, which seems to be a more helpful indicator than relying heavily on default rates. But since repayment rates are a new measure, colleges had no incentive to manipulate repayment rates as they did default rates. This creates a research opportunity to examine whether colleges may have been acting strategically to lower default rates even as their students’ underlying financial situations did not change.

I teamed up with Amy Li, an assistant professor at the University of Northern Colorado, to examine whether the factors affecting loan repayment rates differ from those factors affecting default rates—and whether the factors affecting repayment rates varied based on the number of years after the student entered repayment. Our article on this topic is now out in the ANNALS of the American Academy of Political and Social Science, with a pre-publication version available on my personal website.

We used default and repayment data on students who entered repayment in fiscal years 2006 and 2007 so we could track repayment rates over time. Default rates at the time covered the same time period as the one-year repayment rate, while we also looked at repayment rates three, five, and seven years after entering repayment. (And we had to scramble to redo our analyses this January, as the Department of Education announced a coding error in their repayment rate data in the last week of the Obama Administration that significantly lowered loan repayment rates. If my blog post on the error was particularly scathing, trying to revise this paper during the journal editing process was why!) We then used regressions to see which institutional-level factors were associated with both default and non-repayment rates.

Our key findings were the following:

(1) Being a traditionally underrepresented student was a stronger predictor of non-repayment than default. Higher percentages of first-generation, independent, first-generation, or African-American students were much more strongly associated with not repaying loans than defaulting after controlling for other factors. This suggests that students may be avoiding default (perhaps with some help from their former colleges), but they are struggling to pay down principal soon after leaving college.

(2) For-profit colleges had higher non-repayment rates than default rates. Being a for-profit college (compared to a public college) was associated with a 1.7 percent increase in default rates, yet an 8.5% increase in non-repayment. Given the pressure colleges face to keep default rates below the threshold needed to maintain federal loan eligibility—and the political pressures for-profit colleges have faced—this result strongly suggests that colleges are engaging in default management strategies.

(3) The factors affecting repayment rates changed relatively little in importance over time. Although there were some statistically significant differences in coefficients between one-year and seven-year repayment rates, the general story is that a higher percentage of underrepresented students was associated with higher levels of non-repayment across time.

As loan repayment rates (hopefully!) continue to be reported in the College Scorecard, it will be interesting to see whether colleges try to manipulate that measure by helping students close to repaying $1 in principal get over that threshold. If the factors affecting repayment rates significantly change for students who entered repayment after 2015, that is another powerful indicator that colleges try to look good on performance metrics. On the other hand, the growth of income-driven repayment systems that allow students to be current on their loans without repaying principal, could also change the relationships. In either case, as colleges adapt to a new accountability system, policymakers would be wise to consider additional metrics in order to get a better measure of a college’s true performance.

Posted in Uncategorized | Tagged , , , | 3 Comments

The Challenges Facing New York’s Tuition-Free College Program

Although tuition-free public college will not become a federal policy anytime soon, more states and local communities are considering different variations of free college. There are nearly 200 active college promise or free college programs in the United States, with two states (Arkansas and New York) enacting tuition-free programs in recent weeks.

New York’s Excelsior Scholarship program has garnered quite a bit of attention because it covers students at four-year colleges (most larger programs are limited to less-expensive two-year colleges), because of the conditions attached, and because New York governor Andrew Cuomo is likely to run for president in 2020. Yet the ambitious program (the legislation text starts on page 142 of this .pdf) also has to overcome a number of challenges in order to be truly effective. I discuss three of the key challenges with this program below.

Challenge 1: Will scholarship funds be available to all qualified students? The budget includes $163 million in funding for the program, which is probably far below the amount of money needed to fund all students. Judith Scott-Clayton of Teachers College estimated that an earlier version of the program could cost about $482 million per year. Even requirements that students complete 30 credits per year and clawbacks for students who leave the state after graduation (more on that later) may not bring the cost down enough—particularly if the program is successful in increasing enrollment at public colleges. The budget has a provision that allows awards to be cut or allocated via lottery if funds run short, which is a distinct possibility if the state faces another recession. Needless to say, this would be a PR nightmare for the state.

Challenge 2: Will colleges use fees as a tuition substitute? A full-tuition scholarship sounds great, but students and their families often forget about fees. Right now, fees are a sizable portion of direct educational prices. For example, at SUNY-Albany, tuition is $6,470 and fees are $2,793, while Hostos Community College charges $4,800 in tuition per year for a full-time student alongside $406 in fees. Since the scholarship only covers tuition, the state may pressure colleges to increase fees in an effort to reduce program costs. This happened in Massachusetts for years and still happens in Georgia, both states with large merit-based grant aid programs. Over time, it is quite possible that the value of the grant fails to keep up with inflation as a result—particularly if the state shifts funding from appropriations to student aid and colleges scramble for another revenue source.

Challenge 3: Will the state be able to manage a large “groan” program? Perhaps the most controversial portion of New York’s program is the requirement that students must live and work in the state after college for the same number of years that they received the grant; if they fail to do so, the grant converts to a loan (also known as a “groan” to financial aid wonks). Many people have raised concerns about the fairness of this idea, but here I’ll touch on the logistics of the program. Can the state of New York track students after graduation and see where they both live and work? Will they feel pressures to exempt students who live out of state but work in New York and pay state income taxes? What will the terms of the converted loans look like? There are a lot of unanswered questions here, but it is clear that the state must invest in a larger student loan agency in order to manage this complex of a program.

As Governor Cuomo prepares for a likely presidential bid in 2020, he is counting on the tuition-free college proposal to be one of his signature policy ideas. Some of the biggest concerns with this legislation may take years to develop, but even a period of two or three years may be enough to see whether the program can work effectively around some of the significant concerns noted here.

Posted in Uncategorized | Tagged , , , , | Leave a comment

The Importance of Negative Expected Family Contributions

The Free Application for Federal Student Aid (FAFSA) has received a great deal of attention in the past year. From a much-needed change that allowed students to file the FAFSA in October instead of January for the following academic year to the pulling of the IRS Data Retrieval Tool that made FAFSA filing easier for millions of students, the federal financial aid system has had its ups and downs. But one criticism that has been consistent for years is that the FAFSA remains an extremely blunt—and complex—financial aid allocation instrument.

After students fill out the FAFSA, they receive an expected family contribution (EFC), which determines their eligibility for federal and other types of financial aid. EFCs are currently truncated at zero for reporting purposes, which lumps together millions of students with various levels of (high) financial need into the zero EFC category. In a previous article, I showed that more than one-third of undergraduate students have a zero EFC and how that rate has generally increased over time.

Yet the underlying FAFSA data allows for negative EFCs to be calculated, and these negative EFCs can be used for two different purposes. First, they could be used to give additional Pell Grant aid to the neediest students; there have been several proposals in the past to allow EFCs to go down to -$750 in order to boost Pell Grants by up to $750. Second, the sheer number of students classified in the zero EFC category makes identifying the very neediest students difficult when there are insufficient funds to help all students from lower-income families. Reporting negative EFCs would at least allow colleges to help target their often-scarce resources in the best possible manner.

In my newest article (just published in the Journal of Student Financial Aid, which is open-access!), I used five years of student-level FAFSA data from nine colleges to show how calculating negative EFCs can help identify students with the greatest levels of financial need. The graphics below give a rough idea of what the distributions of negative EFCs could look like under various scenarios and current FAFSA filing situations. (I show dependent students here, but the same story is generally true for independent students.)

I also looked at how much it might cost the federal Pell Grant program to fund EFCs of -$750 by increasing maximum Pell Grants by an additional $750 for the neediest students. I estimated that funding negative EFCs would have increased Pell Grant expenditures by between $5 billion and $7 billion per year, depending on the specification. This is far from a trivial change for a program that spent about $31.5 billion in 2013-14, but it would roughly return Pell spending to its high point following the Great Recession. To save money, additional Pell funds could be given just to students with an automatic zero EFC—students with low family incomes who are already receiving some kind of means-tested benefit (such as free lunches in high school). That sort of limited expansion could be funded out of the current Pell surplus (assuming it doesn’t get used for other purposes, as is currently proposed).

Regardless of whether students get more money from the federal government under a negative EFC, it is a no-brainer for Congress and the Department of Education to work together to at least release the negative EFC number alongside the current number. That way, states, colleges, and private foundations can better target their funds to students with the absolute greatest need. Until the FAFSA is simplified, it makes sense to better use all of the information that is collected on students so everyone can make better decisions on allocating scarce resources.

Posted in Uncategorized | Tagged , , , | Leave a comment

How Popular Was the IRS Data Retrieval Tool?

The financial aid application season for the 2017-18 academic year started out on a high note for current and prospective students. Thanks to the adoption of “prior prior year” or “early FAFSA,” students could file the FAFSA beginning October 1 instead of the following January 1 for the 2017-18 academic year. Students took advantage of this change in large numbers, with about 5.4 million students completing the FAFSA before the previous opening date of January 1.

But FAFSA filing hit a significant roadblock in early March when the federal government quietly pulled access to the IRS Data Retrieval Tool (DRT), which allowed students to quickly and seamlessly transfer their tax records from the IRS to the FAFSA. The tool was down for nearly a week before the IRS issued a statement explaining that the site had been taken offline due to security concerns—and now it looks like the Data Retrieval Tool will be down until next fall at the earliest. Students can still complete the FAFSA by inputting information from their 2015 tax returns, but this is an extra hurdle for many students to jump.

It is possible that the DRT outage is already affecting FAFSA filing rates. Nick Hillman of the University of Wisconsin-Madison (one of the best higher ed finance researchers out there) and his sharp grad students Valerie Crespin-Trujillo and Ellie Bruckner) have been tracking FAFSA filing trends among high school students since the start of this application cycle. Their latest look at filing trends (which they update every Friday) shows the following, which suggests a possible dip due to the DRT outage.

One question that hasn’t been addressed yet is how many students were actually using the DRT when it was pulled. Unlike the great data that Federal Student Aid makes available on FAFSA filing trends, far less data are available on DRT usage. But I was able to find two data points that provide some insights about how many FAFSA filers used the DRT. The first data point came from Federal Student Aid’s 2016 annual financial report, which listed the DRT as a priority for the department. As the highlighted text below shows, about half of all applicants who filed taxes used the DRT in the 2014-15 filing season.

A tidbit of more recent data comes from a presentation that Federal Student Aid made to a conference of financial aid professionals last fall. As shown below, 56% of the 2.2 million FAFSA filers in October 2016 used the DRT. Early FAFSA filers may have different characteristics than filers across the whole application cycle, but this again shows the popularity of the DRT.

Another important group of students use the Data Retrieval Tool—students who are enrolling in income-driven repayment plans. These students have to certify their income on an annual basis (and a majority of borrowers already struggle to do this on time), which becomes more time-consuming without the DRT. It’s still possible for students to do by submitting documentation of income, but the loss of the DRT makes that a lengthier process. I was unable to find any information about DRT usage among people in income-driven repayment programs, but my gut instinct is that it’s a fairly high percentage of borrowers.

The bottom line here—the lengthy outage of the IRS Data Retrieval Tool doesn’t mean that students can’t apply for federal financial aid or income-driven student loan repayment programs. But it does create an additional roadblock for millions of students, their families, and financial aid offices to navigate. Only time will tell whether the DRT outage is associated with lower FAFSA or income-driven repayment filing rates, but a small negative effect seems plausible.

Thanks to Carlo Salerno of Strada Education for inspiring me to dig into the numbers. Twitter conversations can be useful, after all!

Posted in Uncategorized | Tagged , , , | 1 Comment

The 2017 Net Price Madness Bracket

Every year, I take the 68 teams in the NCAA Division I men’s basketball tournament and fill out a bracket based on colleges with the lowest net price of attendance (defined as the total cost of attendance less all grant aid received). My 2016, 2015, 2014 and 2013 brackets are preserved for posterity—and often aren’t terribly successful on the hardwood. My 2015 winner (Wichita State) won two games in the tournament, while prior winners Fresno State (2016), Louisiana-Lafayette (2014), and North Carolina A&T (2013) emerged victorious for having the lowest net price but failed to win a single game. However, North Carolina (a Final Four selection for low-income students in 2016) did actually advance to the championship game before getting beaten by pricey Villanova.

I created two brackets this year using 2014-15 data (the most recent available through the U.S. Department of Education): one for the net price of attendance for all first-time, full-time students receiving grant aid and one focusing on students who received federal financial aid with family incomes below $30,000 per year. I should note that these net price measures are far from perfect—the data are now three years old and colleges can manipulate these numbers through the living allowance portion of the cost of attendance. Nevertheless, they provide some insights regarding college affordability—and they may not be a bad way to pick that tossup 8/9 game that you’ll probably get wrong anyway.

The final four teams in each bracket are the following, with the full dataset available here:

All students receiving grant aid

East: New Orleans ($8,867)

West: West Virginia ($10,405)

South: Northern Kentucky ($9,173)

Midwest: North Carolina Central ($9,793)

Low-income students only

East: Florida ($7,024)

West: Princeton ($3,461)

South: Northern Kentucky (5,030)

Midwest: Michigan ($3,414)

A big congratulations to the University of New Orleans for having the lowest net price for all students and to the University of Michigan for having the lowest net price for its (fairly small percentage of) low-income students. And a hearty lack of congratulations to Southern Methodist for having the highest net price for all students ($36,602) and Gonzaga for having the highest for low-income students ($30,166).

Posted in Uncategorized | Tagged , | Leave a comment

Do Financial Responsibility Scores Predict College Closures?

The U.S. Department of Education’s Office of Federal Student Aid quietly released data on the financial responsibility scores of private nonprofit and for-profit colleges earlier this week, something that they have done for each of the last nine years. These scores, which can range from -1.0 to 3.0, are designed to represent a college’s financial health (although some colleges dispute the value of these scores). A score of 1.5 or above represents a passing score, meaning colleges can receive federal financial aid dollars without any additional restrictions.

Colleges scoring 0.9 or below fail the financial responsibility test and must submit a letter of credit to the Department of Education and submit to additional oversight in order to receive federal funds, while colleges scoring between 1.0 and 1.4 receive additional oversight but do not have to submit a letter of credit. Colleges in the worst financial shape may not even receive a score, as the Department of Education can instead choose to place a college under heightened cash monitoring rules (similar to the penalties for failing) without even doing the calculations.

In the newly-released data for the 2014-15 fiscal year, 187 colleges failed, 139 were in the oversight zone, while 3,048 passed unconditionally. The number of failures is the lowest on record, while the number in the oversight zone is also relatively low compared to past years. But at the same time, the rate of college closures increased sharply last year. Does this mean that financial responsibility scores are identifying financially struggling colleges, or is the metric incorrectly identifying colleges at risk of closure as being financially stable?

To answer this question, I used the best existing database of college closures—from Ray Brown’s College History Garden blog. (Check it out!) I examined the fourteen accredited private nonprofit colleges that closed in 2016 to see what the college’s financial responsibility score was in the 2014-15 fiscal year. (For colleges without a score, I checked the heightened cash monitoring (HCM) list as of September 1, 2015.) The results are below.

Name Score (2014-15)
AIB College of Business  N/A
American Indian College -0.2
Barber-Scotia College  N/A
Burlington College HCM
Colorado Heights University 2.2
Crossroads College HCM
Dowling College 0.6
Kilian Community College 1.8
Northwest Institute of Literary Arts -0.9
Ohio College of Massotherapy 2.7
Saint Catharine College HCM
The Robert B. Miller College -1
Trinity Lutheran College 0.6
Wright Career College 1.1


Two of the 14 colleges did not show up as either having a financial responsibility score or being under HCM, while three other colleges were on HCM due to financial issues and did not receive a financial responsibility score. Of the other nine colleges, four received a passing financial responsibility score (the Ohio College of Massotherapy received the same score as Yale), two were in the additional oversight zone, and three failed. This suggests that either financial conditions changed considerably between mid-2015 and 2016 for some colleges or that financial responsibility scores are an imperfect measure of a college’s fiscal health.

Posted in Uncategorized | Tagged , , | 2 Comments

Examining College Endowments per Pell Recipient

One of the most-discussed higher education policy proposals from President Donald Trump has been a proposal to tax the endowments of wealthy colleges that are seen as not using enough money on financial aid. Key Trump supporter Rep. Tom Reed (R-NY) has introduced legislation requiring colleges with endowments over $1 billion to spend at least 25% of all investment returns on financial aid, much to the chagrin of wealthy colleges.

This proposal does not take into account the size of a college—which means that colleges with similar endowment levels can have vastly different levels of resources. For example, Vassar College and North Carolina State University had endowments just under $1 billion as of June 2015, but the sizes of the institutions are far different. Vassar has about 2,500 undergraduate students, while NC State has nearly ten times as many.

Another important factor is the financial need of students. Colleges can have similar sizes and similar endowment levels, but differ substantially in their number of Pell recipients (a proxy for low-income status). Washington State University and the University of Missouri-Columbia both have endowments around $900 million, but Washington State enrolled 3,000 more Pell recipients than Mizzou in spite of enrolling 4,000 fewer undergraduates. This means that Mizzou has the ability to target more aid to their Pell recipients should they choose to do so.

To explore this point in more detail (and thanks to Sara Goldrick-Rab for the idea), I dove into newly available finance data from the Integrated Postsecondary Education Data System (IPEDS) for the 2014-15 academic year and merged it with data on the number of Pell recipients for the same year from Federal Student Aid’s Title IV volume report datasets. After eliminating colleges that did not report endowment values or reported endowment or Pell recipient data in conjunction with other campuses, my sample consisted of 479 public four-year colleges and 909 private nonprofit colleges. You can download the spreadsheet here to see the ratios for each college with data. (Note: This was updated on February 20 to include colleges in the District of Columbia. Thanks to Patricia McGuire for calling that error to my attention!)

Most colleges have quite small endowments per Pell recipient, as shown in the graph below. The median public college had an endowment of $12,778 per Pell recipient in 2014-15, while the median private college had an endowment of $65,295. Given typical endowment spending rates of about 5% per year, this means that public colleges can spend about $640 per Pell recipient on financial aid and private colleges could spend about $3,200 per recipient. But this assumes that (1) colleges will only spend their endowment proceeds on need-based aid and (2) colleges can actually use their endowments on whatever they want instead of what donors say. This means that most colleges do not have much ability to significantly improve financial aid packages based on endowment proceeds alone.


The other thing that stands out in the graph is the number of colleges with endowments of over $1 million per Pell recipient. In 2014-15, 92 colleges were in the millionaires’ club, including 88 private nonprofit colleges and four public colleges (William and Mary, Michigan, Virginia, and Virginia Military Institute—an unusual institution). Below are the institutions with the 25 highest endowment to Pell ratios. All of these colleges have more than $4.2 million per Pell recipient—an enviable position should any of these colleges seek to increase low-income student enrollment.

Name Undergrad enrollment Pell enrollment Endowment ($bil) Endowment per Pell recipient ($mil)
Johns Hopkins University 6357 787 3.33 4.23
Grinnell College 1734 393 1.79 4.55
Claremont McKenna College 1301 152 0.73 4.83
Amherst College 1792 442 2.19 4.96
Bowdoin College 1805 278 1.39 5.01
Columbia University in the City of New York 8100 1912 9.64 5.04
Williams College 2072 403 2.27 5.62
Northwestern University 9048 1256 7.59 6.04
University of Pennsylvania 11548 1643 10.10 6.17
Pomona College 1650 326 2.10 6.44
Swarthmore College 1542 237 1.85 7.79
Dartmouth College 4289 596 4.66 7.82
Duke University 6626 925 7.30 7.89
Washington and Lee University 1890 181 1.47 8.13
Rice University 3926 620 5.57 8.99
University of Notre Dame 8448 902 8.78 9.74
Soka University of America 411 123 1.22 9.93
University of Chicago 5738 639 6.55 10.30
Washington University in St Louis 7401 571 6.89 12.10
California Institute of Technology 983 127 2.08 16.40
Massachusetts Institute of Technology 4512 806 13.50 16.70
Stanford University 7019 1106 22.20 20.10
Princeton University 5391 790 22.30 28.20
Harvard University 10338 1238 37.60 30.40
Yale University 5477 724 25.50 35.30
Posted in Uncategorized | Tagged , , ,

Should Part-Time Students Have Their Borrowing Limited?

One of the key higher education policy interests of Senate Health, Education, Labor, and Pensions chairman Lamar Alexander (R-TN) has been to limit student borrowing in an effort to help reduce rising student loan debt. I’ve written in the past about how “overborrowing” is not as big of a concern as students not borrowing enough for college, but there is one group of students that may actually benefit from not being able to take out the maximum allowed amount in student loans.

Currently, students who attend college part-time can borrow the same amount as full-time students as long as there is space in their financial aid package. This can be a concern for students, as it means that they can run out of federal loan eligibility before they complete a bachelor’s degree. Current federal loan limits are the following:

Year in college Dependent student Independent student
First $5,500 $9,500
Second $6,500 $10,500
Third $7,500 $12,500
Fourth and beyond $7,500 $12,500
Lifetime $31,000 $57,500


This equates to about four and a half years of borrowing at the maximum for dependent students and five years for independent students. Given that a sizable percentage of students complete a bachelor’s degree in more than five years, running out of loan eligibility before graduation can be a real concern for students. This is particularly true among students who begin at a community college, where tuition is relatively low compared to at a four-year college. If a student reasonably expects to take six years to complete a bachelor’s degree, then she and her financial aid office should have a conversation about how to best preserve her loan eligibility for when she needs it the most.

A fairly straightforward way to reduce the number of students who exhaust their loan eligibility would be to allow students to get a certain amount of money per credit hour. Students can currently receive a Pell Grant for up to 12 full-time equivalent semesters, with full-time defined as taking at least 12 credits. The current loan limit could be divided by 12 (roughly $2,600 per semester), or this could be done on a per-credit basis (perhaps $200 per credit) to recognize that students who take more classes need to work less.

A completely different proposal would allow students to use their student loan eligibility in any way they see fit. For example, dependent students could use their $31,000 in two years if desired—as long as they had space in their aid package. This idea of an education line of credit was raised by Jeb Bush in his short-lived presidential campaign, but it is unclear what Senator Alexander thinks of this proposal. At this point, it seems like the idea of limiting borrowing for part-time students at an individual college’s discretion is the most likely policy outcome.

Posted in Uncategorized | Tagged , , , | 2 Comments

How Much Did A Coding Error Affect Student Loan Repayment Rates?

Mistakes happen. I should know—I make more than my fair share of them (including on this blog). But some mistakes are a little more noticeable than others, such as when your mistake has been viewed more than a million times. That is what happened to the U.S. Department of Education recently, when they found a coding error in the popular College Scorecard website and dataset.

Here is a description of the coding error from the Department of Education’s announcement:

“Repayment rates measure the percentage of undergraduate borrowers who have not defaulted and who have repaid at least one dollar of their principal balance over a certain period of time (1, 3, 5, or 7 years after entering repayment). An error in the original college scorecard coding to calculate repayment rates led to the undercounting of some borrowers who had not reduced their loan balances by at least one dollar, and therefore inflated repayment rates for most institutions. The relative difference—that is, whether an institution fell above, about, or below average—was modest.  Over 90 percent of institutions on the College Scorecard tool did not change categories (i.e., above, about, or below average) from the previously published rates. However, in some cases, the nominal differences were significant.”

As soon as I learned about the error, I immediately started digging in to see how much it affected loan repayment rates. After both my trusty computer and I made a lot of noise trying to process the large files in a short period of time, I was able to come up with some top-level results. It turns out that the changes in loan repayment rates are very large. Three-year repayment rates fell from 61% to 41%, five-year repayment rates fell from 61% to 47%, and seven-year repayment rates fell from 66% to 57%. These changes were quite similar across sectors.


Difference between corrected and previous loan repayment rates (pct).
Corrected Previous Difference N
All colleges
  3-year 41.0 61.0 -20.0 6,090
  5-year 47.1 61.1 -14.0 5,842
  7-year 56.7 66.3 -9.6 5,621
  3-year 46.6 66.8 -20.2 1,646
  5-year 54.2 68.9 -14.7 1,600
  7-year 62.1 72.1 -10.0 1,565
Private nonprofit
  3-year 57.7 77.5 -19.8 1,386
  5-year 63.7 77.3 -13.6 1,375
  7-year 70.4 79.3 -8.9 1,338
  3-year 30.5 50.4 -19.9 3,058
  5-year 35.0 48.9 -13.9 2,850
  7-year 46.9 56.5 -9.6 2,700
Source: College Scorecard.


For those who wish to dig into individual colleges’ repayment rates, here is a spreadsheet of the new and old 3, 5, and 7-year repayment rates.

Fixing the coding error made a big difference in the percentage of students who are making at least some progress repaying their loans. (And ED’s announcement yesterday that it will create a public microdata file from the National Student Loan Data System will help make these errors less likely in the future as researchers spot discrepancies.) This change is likely to get a lot of discussion in coming days, particularly as the new Congress and the incoming Trump administration get ready to consider potential changes to the federal student loan system.

Posted in Uncategorized | Tagged , , , | 9 Comments