Chapter 14: Data & Analytics
IN THIS LESSON
What is a custom score model?
A scorecard (or model) is a mathematical function that is developed using statistical techniques and generates a score. Significant variables (or characteristics) are methodically chosen and assigned a weight. The variable weights are summed to create a final score. The score is a representation of the event that is being predicted.
A custom scorecard is used very similarly to the way a generic score (such as FICO® score or Vantage® Score) is used. However, a custom scorecard gives more power to differentiate levels of risk in the underwriting process (decisioning optimization), setting up loan structures, and defining the right level of pricing (risk-based pricing). This power comes from being able to identify characteristics of the consumer base that are unique to the financial institution’s portfolio and its population. In addition, custom scorecards typically use application information and loan details that are predictive but not utilized within generic scorecards.
Benefits of a custom model/scorecard
Custom scorecards
optimize risk levels
allow for increased decisioning efficiency and automation
provide opportunities for business growth
Because the custom score is more predictive than the generic score, it will drive more of those potential bad loans into lower score ranges (more auto-decline) and more expected good loans into higher scores (auto-approval). It will allow the financial institutions to evaluate the true risk of some applicants that would have been otherwise misjudged through a generic score (more business and less loss). The generic score may accomplish this to some extent, but a custom score typically would perform much better.
Institutions using custom scorecards can free resources to focus on the review, evaluation, and decisioning of applications that may need more scrutiny. At the same time, with the more precise risk levels associated with a custom scorecard, loan pricing will be more accurate because it is better aligned with the more accurate projected loan risk.