Point Predictive tool automates car loan pricing, fraud protection

A new Point Predictive tool allows auto lenders to automatically reject an applicant because of the likelihood of fraud or set interest rates to reflect that risk, according to the company. Paired with a new capability to handle inquiries from denied customers, the new system is compliant with federal law, Point Predictive said.

The AutoPass system announced last summer increases the pool of loan applications that can be "auto-decisioned" — approved or rejected immediately by a lender's software without human review. This gives a dealership and customer a speedier answer.

Point Predictive's earlier Auto Fraud Manager screening system featured software that alerted lenders to potential fraudsters, but only allowed them to deal with suspicious applicants in a "semiautomated fashion," according to Point Predictive Chief Fraud Strategist Frank McKenna. The bank would have to ask the customer for documentation and then base an approval or denial based on that, he said.

"They can't decline a consumer outright," McKenna said.

Basing a decision on the software's modeling would leave a lender out of compliance with the law, he said. But the additional work required for compliance means a lender could be wasting time and resources on an applicant it views as probably running a scam.

McKenna said the new AutoPass tool allows a lender to skip this step, save "quite a bit" of expense and automate the application.

The new system expands Point Predictive's role in the process when a loan is automatically declined as a fraud risk.

A lender must tell a borrower why their application was denied, and a customer must be given the opportunity to challenge the decision, McKenna said. In this case, they'd contact Point Predictive, and McKenna said his fraud analytics company had to invest in the infrastructure to receive these customers' inquiries.

"We have to be available to take those disputes from customers," McKenna said.

McKenna said the new product wasn't facilitated by a change in federal law; such fraud-related auto-decisioning had already been permitted. Point Predictive made the decision to pursue this capability after observing increased interest from lenders as that

AutoPass could enable automated decisions for as many as 80 percent of applications — double where some lenders sit today, McKenna said.

McKenna said Point Predictive undertook many modeling exercises to ensure compliance and spent considerable effort developing, training and validating the system's model. He called the end result's accuracy "on par" with its traditional Auto Fraud Manager fraud scoring model, which was anticipated to retain a significant customer base. Some lenders would prefer to continue checking for fraud but retaining practices like stipulations, he said.

McKenna said the AutoPass system also can be used by lenders to price loans based upon the fraud risk perceived and stay compliant, another service new to Point Predictive.

A customer flagged by the system as a high fraud risk would likely be automatically declined, McKenna said. But a lender might approve an edge case, just with a higher interest rate to make up for the potential fraud threat, he said.

McKenna gave the example of a lender's thinking when they suspect minor income misrepresentation.

"'I don't want to go through all the process of looking at all the [stipulation] paperwork. I'm just gonna price it in,'" he said.

This cuts both ways. McKenna predicted lenders would shave off interest on borrowers flagged as a lower fraud risk, and he said this was the more likely pricing scenario to result from AutoPass.

"I think that's where a lion's share of the benefit might be is actually in the lowest [fraud] scores and maybe more positive pricing, versus puntitive pricing," he said.