Fully explainable machine learning model enables Prestige to double its loan volume in a six-month period with no added risk.
ZestFinance, a leading artificial intelligence (AI) software company that uses machine learning to vastly improve credit underwriting, today announced a partnership with subprime auto lender Prestige Financial Services to implement the industry’s first fully explainable machine learning model to predict borrower risk. Prestige replaced legacy scoring techniques with Zest Automated Machine Learning (ZAML) software, allowing the lender to harness thousands of data variables and advanced math in its credit models.
The new models enabled Prestige to achieve a 36 percent increase in new applicants, resulting in a 14 percent higher approval rate. Within six months of deploying the AI-based credit underwriting model, Prestige doubled its lending volume without added portfolio risk.
Prestige is bucking the trend in the subprime auto lending sector, which is seeing rising defaults and tighter underwriting standards to control losses. At the time Prestige engaged ZestFinance, the company had raised its underwriting thresholds to the point where roughly seven out of ten applicants were turned down for a loan. Applying machine learning enabled Prestige to rank-order risk more effectively across all types of borrowers, allowing Prestige to swap out risky borrowers and replace them with thin-file and new-to-credit consumers who were more creditworthy than a traditional credit score might suggest.
“We had considered using machine learning models in the past,” said Steven Warnick, chief credit and analytics officer at Prestige. “We knew they were better at predicting risk, but had concerns because we couldn’t explain them. ZestFinance not only helped us build better models of predictability, but also provided a read-out of the key factors that led to the credit model’s output, so we can fully comply with all regulatory requirements. That was the game changer.”
“It’s been shown over time that AI-based underwriting can help financial service organizations better evaluate the creditworthiness of applicants, expanding credit options to applicants who might not otherwise qualify using traditional credit scoring,” said ZestFinance founder and CEO, Douglas Merrill. “But the historic lack of explainability and model-risk management tools has made lenders reluctant to transition to machine learning technology. We’ve focused on building technologies that provide clear and comprehensive explanations for every underwriting decision.”
The work with ZestFinance has allowed Prestige to achieve its original goal of approving more borrowers without taking on more risk. It’s also resulted in greater interest from independent car dealerships across the country. As a result of the initial success, Prestige is developing plans for other ways to automate and innovate in traditional auto lending. Download the case study here.
