The Evolution of Risk Scoring

The Sooner, the Better: How Risk Scoring Became a Powerful Pre-Repo Tool

Guest Article

DRN data has long been used in collections and repossessions to find more vehicles faster. We believed this data could be just as powerful for lenders as they scored loans and considered when to assign a vehicle for repossession. Our idea: use DRN’s vehicle location data and analytics to assess risk and influence decisioning after origination and before repossession.

Our idea: use DRN’s vehicle location data and analytics to assess risk and influence decisioning after origination and before repossession. Lenders use traditional attributes such as missed payments, numbers of loan extensions and FICO score trends in their behavior and repo scoring models to determine when a vehicle should be repossessed.

DRN always believed data attributes like scans at an address or scans a specified distance away from the given address could help tell a story and provide useful insights into a borrower’s behavior that might accelerate repo timing. Or, the data may help the lender delay repossession and collect on a vehicle if it shows a consistent address that leads to contact.

In either case, Risk Scoring became a solution that our team developed and discussed with lenders for years before putting it into play. The game changer? Machine learning. Until 2019, validating the state name on a license plate was done manually – which took lots of manpower to work through hundreds of thousands of records at a time. Everything changed when our IT group developed machine learning to take over this task. This not only makes state identification faster, but also more precise. For example, a file of 250K VINs would take 3 months to manually validate all scans with 90% accuracy. With machine learning, it takes only hours with up to 99% accuracy.

The bottom line: better decisions, lower risk, improved repo timing or avoiding repo altogether by making contact. Risk Scoring does all of this for lenders, and the sooner, the better.