FICO: Understanding the Risks of Multiple Credit Scores in Mortgage Lending

(This article, by Clifford Rossi of the University of Maryland, was submitted by FICO, San Jose, Calif.)

Clifford Rossi is an Executive-in-Residence and Professor of the Practice at the Robert H. Smith School of Business, University of Maryland. Prior to entering academia, Rossi had nearly 25 years of experience in banking and government, having held senior executive roles in risk management at several of the largest financial services companies.

As the Federal Housing Finance Agency considers whether to update the credit scoring models used by Fannie Mae and Freddie Mac, a new paper from Dr. Clifford Rossi explores the risks of introducing multiple credit scores to the mortgage lending system. Specifically, Rossi examines “the effects of alternative credit scoring models on credit risk, profitability and effectiveness at expanding access to credit,” using three hypothetical mortgage score models and a large sample of GSE loans originated between 1999 and 2015.

Rossi is one of the country’s leading risk management experts, having held numerous senior executive roles in risk management at several of the nation’s largest financial services companies. He was on the team at Freddie Mac that developed the industry’s first automated underwriting system and personally developed the first FHA and VA statistically based underwriting scorecards.

Among the key findings from his paper:

• Scoring models from different developers don’t represent equivalent credit risk: Rossi finds that a model that reduces minimum scoring criteria performs “demonstrably weaker” than a model that replicates the FICO Score. He concludes that “these differences in performance result in significant variations in credit risk … in other words, a score of 660 [across the two models] does not translate into the same risk.” 

• Mandating interchangeability between non-comparable scoring models could increase risk across the mortgage system — without expanding credit access: Because different models are not equivalent, a system in which lenders select between multiple scoring models could end up increasing risk (and thus costs) across the entire mortgage market, Rossi writes. He notes that “alignment of credit risk across different credit scores is possible at a point in time but can deteriorate over time and with varying economic conditions.” What’s more, scores that reduce minimum scoring criteria “may not result in a significant uptake in acceptable loans.”

• Segments with sparse data could have negative profitability. Those who are scorable despite sparse data are fundamentally different than those with robust data. Scores for those with sparse data will have weaker predictive performance and a 660 is likely to be significantly riskier for this segment than a 660 in other segments in different economic cycles. Score cutoffs will be determined by the entire population, and not this segment, which could lead to negative profitability for this group.

• The FICO Score has been a reliable predictor of mortgage risk for decades: Rossi notes that “FICO scores leverage detailed credit information on individuals from the major credit repositories to model credit delinquency” and finds that the FICO Score model currently used by the GSEs “has been proven to be highly significant from a statistical perspective over time when modeling” a consumer’s likelihood of mortgage repayment.

Read the entire research study by clicking here.

(Views expressed in this article do not necessarily reflect policy of the Mortgage Bankers Association, nor do they connote an MBA endorsement of a specific company, product or service. MBA NewsLink welcomes your submissions. Inquiries can be sent to Mike Sorohan, editor, at msorohan@mba.org; or Michael Tucker, editorial manager, at mtucker@mba.org.)