Introduction
On 18 January 2017, the Consumer Financial Protection Bureau (“Bureau” or CFPB) filed a lawsuit against a large student loan servicer that manages over $300 billion in federal and private loans for over 12 million borrowers. The CFPB alleged that the servicer failed to provide borrowers with adequate information about repayment plans when borrowers were unable to pay and did not help them identify eligibility for and enrollment in income-based repayment plans.
In this litigation, the CFPB relied on the number of complaints that borrowers and co-signers filed with the Bureau in its Consumer Complaint Database (“Complaint Database”). The CFPB focuses exclusively on the number of complaints when ranking student loan servicers in its Annual Reports of the Student Loan Ombudsman.2 Academic sources note that the Bureau uses the Complaint Database to inform rulemaking, enforcement actions, and consumer education.3 The database helps the CFPB process consumer complaints and provides some information about problems that consumers face. However, there are statistical and economic limitations that may render the number of complaints or the complaint rates as unreliable for the purpose of evaluating whether a company is appropriately conducting business.4 The complaint rate is the ratio of borrowers and co-signers of student loans that have filed complaints with the Bureau against a servicer to the total number of borrowers serviced.5
This article illustrates potential limitations of relying on the volume or rate of complaints that borrowers and co-signers of student loans filed with the Bureau and are reported in the public Complaint Database as a basis for the CFPB’s regulatory actions against servicers of student loans and/or to develop servicer rankings in reports used to educate consumers.6
The most notable limitation is self-selection bias. Self-selection bias arises when the individuals that select themselves into a group—such as borrowers who decide to file a complaint with the Bureau—are different, by virtue of their inclusion, from the remaining individuals.7 The Bureau’s complaint database disregards neutral or positive opinions. In addition, each servicer faces a different portfolio of borrowers with different financial situations and ability to repay and a spike in the complaints against a given servicer may reflect the high credit risk of its portfolio rather than the quality of its servicing. Also, the propensity to file a complaint could be influenced by other factors such as negative media coverage, for example. As a result, other factors that could cause the increase in the complaint rate against a given servicer cannot be ruled out if one relied on the Complaint Database to assess servicing quality.
The CFPB does not determine the merits of the complaints and therefore the complaints are an aggregated set of allegations—that may arise from issues outside the control of the servicer—from a self-selected group of borrowers:
… many of the complaints about both federal and private education loans seem to be about issues that are beyond the control of the lender or servicer. We have seen a substantial number of complaints that express generalized unhappiness and frustration with the consumer’s situation—the inability to get a job coupled with concern that the education that they received was not worth the price that they paid and the amount that they borrowed.8
Complaint rates, if used exclusively and without consideration of additional metrics, need not provide a reliable measure of servicer quality because the database simply reflects allegations by dissatisfied borrowers and does not represent the views from a servicer’s entire portfolio.
The CFPB Complaint Database does not include information on the borrowers’ credit attributes, ability to pay, employment, income, loan terms, or other characteristics of the servicers’ loan portfolios. As such, it cannot be used to reliably compare the quality of servicing across servicers given the distinct composition of each servicer’s loan portfolio. Each servicer faces a multitude of individual borrower risk profiles, which may explain the observed discrepancies in complaint rates. Without controlling for the credit attributes of the loans across servicers, complaint rates against different servicers could be an apple-to-orange comparison.9
In fact, our data analysis shows that while student loan servicers experienced notable differences in complaint rates based on the Bureau’s database, the same servicers had similar borrower satisfaction scores according to Department of Education surveys (“DoE surveys”).10 These DoE surveys include metrics to assess consumer satisfaction—not only complaints—and account for variations in each servicer’s loan portfolio. As such, the DoE surveys provide contradictory findings to the CFPB’s Complaint Database regarding borrowers’ satisfaction.11
Another limitation of relying on the complaints filed with the CFPB to evaluate quality of servicing is that the number or the rate of complaints as a measure of quality does not consider the disposition of the complaints.12 Our analysis of the outcomes of private student loan complaints shows that between 73% and 91% are resolved without monetary or non-monetary relief to the consumer each year. Complaint rates may not be a reliable indicator of the quality of loan servicing, given that the CFPB does not validate the merits of the complaints in its database and the majority of the complaints were resolved without the servicer providing the consumer either monetary or non-monetary relief.13
Further, between 77% and 85% of the borrowers did not dispute the resolution of the complaints even for servicers who experienced higher rates of federal loan complaints relative to the other federal loan servicers.14 Indeed, the servicer with the least number of complaints per thousand borrowers was the one with the highest rate of disputed resolutions. While the complaint resolution process can be viewed as another measure of servicer quality, the Bureau’s Annual Reports of the Student Loan Ombudsman focus exclusively on the number of complaints when ranking student loan servicers.15
Lastly, the Bureau uses the Complaint Database to identify trends in the marketplace.16 However, changes in complaint rates over time may be associated with factors like negative press, which can affect the propensity to file complaints. For example, there was a dramatic increase in the complaint rate against student loans servicers immediately after the CFPB’s recent lawsuit in January 2017. On the other hand, borrower satisfaction scores increased slightly after the lawsuit according to the DoE surveys.
In summary, using the CFPB Complaint Database could result in unreliable conclusions about the quality of servicing of student loans due to self-selection bias and the lack of information about the servicers’ loan portfolios, such as loan types and income of borrowers.17 Our empirical evidence shows that the majority of private student loan complaints are resolved without the servicer providing the consumer with monetary or non-monetary relief and the rate of resolving complaints without further disputes is high and similar across servicers. This suggests that complaint rates, by themselves, may not be a reliable indicator of servicer conduct. Lastly, confounding factors such as news coverage may also affect complaint rates, which would further complicate the use of the Complaint Database to assess changes in the quality of servicing over time.
Alternative measures of the quality of student loan servicing may be more reliable. For example, one could compare the loan performance of each servicer’s portfolio after controlling for the credit attributes of the loans and borrowers, such as re-default rates across comparable loan types.18 In addition, various statistical metrics can be used to assess the servicers’ efforts to rehabilitate and restore delinquent or defaulted loans to a paying status, which can be tracked over time rather than at a single point in time. The quality of servicing measures may also include the percentage of borrowers seeking forbearance, deferment, or income-based repayment; the rates at which these repayment inquiries are approved; and whether incomebased repayment plan reminders are initiated by the servicer.
The Student Loans Market
There was approximately $1.3 trillion in outstanding federal loans by the end of 2016.19 In addition, there was an estimated $108 billion in outstanding private student loans.20 In total, outstanding student loan debt at the end of 2016 was approximately $1.4 trillion dollars, as shown in Figure 1.
Federal loans are currently originated under the Direct Loan Program or the Perkins Loan Program.21 For both programs, the borrower’s college, university, or trade school determines his/her eligibility and loan amounts.22 The main difference between the two programs is the lender to whom repayment is owed.23 Under the Direct Loan Program, students receive the loans from and owe repayment to the Department of Education. Under the Perkins Loan Program, schools make loans to students with exceptional financial need and repayment is owed to the schools.
The major differences between federal and private loans are the types of interest rates, repayment options, and whether they are guaranteed by the DoE.24 Interest rates on all federal loans are fixed, and vary by loan type and the date of first disbursement. Federal loans that were credited to borrowers between 1 July 2016 and 30 June 2017 carried fixed interest rates between 3.76% and 6.31%.25 By comparison, private student loans can have either a fixed or an adjustable rate. These interest rates vary by lender and the applicant’s credit application; fixed-rate loans for undergraduates can range between 4.751% and 12.99%, and adjustablerate loans can range between 2.751% and 11.22%.26 Additionally, nearly 90% of private loans require a co-signer.27