In a recent post, I broke down a simplified origination decision model to illustrate how it can be used to improve decision strategies. But many of our clients are beginning to look to more advanced uses of these analytics, to boost originations performance. These lenders are leveraging custom decision models, which are built from their own data, and can incorporate a wide range of profit drivers and relationships.
For instance, lenders can use marketing segments as data inputs to an origination decision model. Advanced marketing segmentation analytics identify highly differentiated segments of customers who are similar in how they respond to offer attributes, channel, pricing and other potential lender actions. When such marketing insights are built into the origination decision model, the optimized decision strategy will more accurately pinpoint the most profitable offer to make to each individual. Lenders can improve not only accept/decline decision strategies, but pricing, credit line, term and condition decisions as well.
One case in point is a large US bank that FICO worked with to simultaneously optimize APR (annual percentage rate) and ICL (initial credit line) decisions for its consumer credit card portfolio. The bank discovered that it could significantly increase profits without increasing exposure. The final strategy makes the most of limited exposure by shifting it from low-risk/low-revenue consumers to low-risk/moderate-revenue consumers. At the same time, it gives higher APRs to riskier accounts with less revenue potential (profiles indicating lower balances and utilization). Expected results include a 17% decrease in loss while maintaining revenue, resulting in a $20 increase in profit per account.
Another FICO client, a large UK bank, was able to increase profit per application by 45%. The objective was to maximize lifetime profit for loans to applicants with no previous relationship with the bank. Many origination decision actions were modeled: accept/decline, loan amount and risk-based pricing—in all, 92 decision combinations. A wide range of business constraints at both the portfolio and account level were used, and a series of "what if?" scenarios were explored. The bank selected an optimal operating point that increased its loss per loan by only 1% for a predicted 46% increase in profit, almost all of which was achieved in the first 16 months of deployment.
Lenders like these are also beginning to model and optimize originations decisions at the customer level, to be able to answer very complex questions like: "Given the most profitable overall credit amount for this customer across all accounts, what size credit line should we assign for this new account?" Lenders who choose origination solutions built on architectures that facilitate sharing of data, business rules, analytic models and treatment outcomes, will be able to move even faster and more cost-effectively to gain competitive edge.
