Applying technology to an all-too-common financial barrier for students.

GUEST COLUMN | by Barbara Freeman and Reginald Berry

Federal student loan programs were founded on the premise of access and equity.

The aim: to enable young people to finance their education in order that they can pursue their life goals and dreams, irrespective of their socio-demographic or financial background.

Yet many young students study and hopefully graduate from postsecondary education with high levels of debt and financial insecurity.

The model, which could help student borrowers avoid economic distress by placing them on the right path when the student loan is initiated, is aimed at supporting students dynamically throughout the student loan process.

This places stress and strain on their lives and can diminish their ability to compete on an equal footing with others less encumbered with debt.

The Struggle is Real

Student loan defaults are substantially higher than defaults on mortgage, auto, and small business loans, and credit card debt.

Borrowers struggle with student loan repayments for many reasons. Some are macroeconomic: escalating tuition costs, slow and disparate regional economic growth, and a difficult job market for entry-level positions.

Other explanations relate to low levels of financial literacy or borrowers’ attitudes and decisions about how to handle student loans.

Graduates tend to prioritize repaying car loans and credit cards over student loans because they are more tangible.

Many students do not understand the complicated student loan options or what it means to default.

Others do not comprehend the growth in the size of loan balances during the college years or fail to realize that they may struggle to secure a well-paying job after graduation.

Still others do not know that their loan must be repaid if they drop out of school.

A Range of Circumstances

Numerous studies enable us to readily identify the range of circumstances that most commonly affect the probability of students’ ability to repay their loans and which factors are most likely to lead to default. The data is robust and can be captured in a few bullet points.

Default is most common with students:

• who drop out of college and fail to complete their degree;
• who complete only a certificate;
• who have small loan balances—students with balances of $5,000 or less are most at-risk of default and student with balances of less than $10,000 account for 65% of all defaults;
• who attend school part-time or on a path that will take them more than the typical four to six years to graduate;
• who attend a for-profit institution (both two and four year) and are the first in their families to get a college degree, typically from a family with income of $40,000 or less;
• who attend an institution for higher education that is ineligible for Title IV funding; and
• who receive Pell grants as opposed to other means of financing.

The federal government possesses massive data that can be used to monitor, and potentially mitigate, these risks.

The Data is Already There

Publicly-available data can be drawn from the US Department of Education Financial Student Aid’s National Student Loan Data (NSLD) and Cohort Default Rate (CDR), the National Postsecondary Student Aid Study, Free Application for Federal Student Aid (FAFSA) data on personal characteristics, and earnings data from the Internal Revenue Service (IRS) and Social Security Administrative (SSA) databases.

Although these data are highly reliable and provide strong analytic value, some data are static and lagging, and individual loan transaction-level details and granular loan performance data are proprietary to the loan servicing companies.

A bipartisan bill, The College Transparency Act of 2017, has been introduced that has the potential to address these issues. It calls for the strengthening of transparency in federal reporting and accessibility of unit-level outcome data (e.g., enrollment, retention, completion, and post-college outcomes).

Even without policy change, federal data can be supplemented by drawing on alternative data points, such as students’ phone and utility records, social media, or psychometrics assessments.

Why Wait?

But why wait to put these available data to work while substantial progress could be made in the near term?

The Student Credit Intervention (SCI; pronounced ‘sky’) is an educational credit intervention for students in the early stage of development.

SCI is a data-driven, predictive model that analyzes borrowers’ attributes and decision-making patterns, identifies critical points at which default is most likely, and uses this data to provide targeted, relevant, and ongoing support to borrowers.

The model, which could help student borrowers avoid economic distress by placing them on the right path when the student loan is initiated, is aimed at supporting students dynamically throughout the student loan process.

Similar to Credit Scoring

The SCI draws on modeling techniques similar to those used in credit risk scoring models.

Models that most of us know, if not by name (FICO, Experian, etc.) then by the fact that they determine whether we qualify for a loan or a mortgage and the terms of the loan.

Typically, credit risk scoring models predict the probability of default within a two-year period or less; creating bands correlated with specific short-term default probabilities. Similar to credit models, the SCI model segments the borrower population and uses a short-term indicator of student loan default.

You may wonder, “Why use a short-term measure when student loans are long-term debt (or more aptly long-term investments)?”

The reason is that short-term predictors provide important information that can be used as a basis for determining the appropriate short-term action or intervention that can help students prevent longer-term problems (see Figure 1 below).

The SCI Model

The SCI model is comprised of a series of individual models, grouped along a dimension according to known risk factors (e.g., financial stressors, academic difficulty). Key Risk Indicators serve as an early warning signal of increasing risk exposure, as exemplified in the following table.

Like credit scoring models, SCI also provides a simple numeric scorecard; except scoring is based on student risk factors and the Key Risk Indicators are designed to trigger alerts to help the student borrower avoid distress.

Thresholds

A key task is to determine the optimal cut-off thresholds for the SCI scoring model. Using these thresholds, the SCI can auto-generate recommendations based on each student’s needs and provide guidance based on the estimated likelihood of various events occurring.

Targeted, relevant, and ongoing support can then be provided to borrowers in order to help them better understand the decisions that they need to make at a given point in time, based on their actions and circumstances.

Warning Signs

Students who are showing warning signs that they may be headed toward trouble can receive greater attention and be provided with the type of information that could lead them down a manageable path.

For example, a text could be sent if an indicator was triggered for a late payment; or if a student is dropping classes, a student could receive a message that helps them understand that additional interest will be accrued (the loan will cost more!) as a result of extending the amount of time until graduation.

In more serious cases, a counselor can be assigned to provide guidance.

Real Help for Students

SCI could also be used at the start of the loan process to better inform loan originations, enabling the government and other lenders to provide borrowers with a fair and appropriate loan product and terms (e.g., flexible income-driven repayment plans), which could be customized to the student’s individual needs rather than providing them with a one-size-fits-all product.

Figure 1 (above) provides an illustration of the SCI Scorecard, working in conjunction with existing credit information.

Sufficient information can be obtained today from the FAFSA and other available data to create a first generation SCI model.

Student borrowers clearly need help.

So, why wait to help students when we can do something now?

—

Barbara Freeman is a Visiting Scholar at the University of California at Berkeley and a consultant to the World Bank. She co-founded KPMG Consulting’s Risk Management practice in the Asia Pacific region and is the co-creator of multiple educational interventions. This article was co-authored by Reginald Berry, who helped develop the federal sector business for FICO and has worked with the SBA, EXIM Bank and Treasury regarding credit risk issues. Special thanks to Doug Criscitello and Kyle Shohfi from the MIT Golub Center for Finance and Policy for their helpful comments. Contact Barbara through LinkedIn.