Lenders face a myriad of challenges these days. To build a profitable portfolio while also managing risk, you need solutions that enable you to streamline decision making, reduce losses, lower costs, approve more profitable applicants, and guarantee regulatory compliance. Bureau scores may be the traditional “tried-and-true” method for assessing customer risk, but they aren’t enough on their own. Using a pooled model in addition to bureau scores can help creditors make more precise, value-based decisions at the origination stage.
What Is a Pooled Model?
A pooled model is a scoring model built on “pools” of historical data from many financial institutions. It is typically viewed as the most practical and cost-effective way to generate scores for new applicants as well as existing customers.
Pooled models can be licensed as “off-the-shelf” and quickly used. This makes them an ideal solution for banks and credit unions that don’t have enough data to create their own custom models but still want the flexibility to grow their portfolios responsibly.
In terms of performance, pooled models tend to be more predictive than bureau scores because they are built on development samples of “similar businesses” within your industry.
Why Using a Bureau Score by Itself May Not Be Sufficient – What’s Missing?
Bureau scores cast a wide net. They identify risk associated with the entire customer credit profile, which is a valuable element when making an origination decision. However, it is not the only metric to use when evaluating risk for new and existing customers.
Note that a bureau score is based on the entire population, including credit seekers and non-credit seekers. Because credit seekers are a small population relative to the entire population, the bureau score is heavily weighted by repayment behavior from non-credit seekers.
An originations score (ARM) is developed solely for credit applicants, making it more sensitive to repayment risk of a new financial obligation. The FICO® ARM Score offers a different view of the same customer to indicate the risk of the specific financial obligation being originated. It provides better separation of “goods” and “bads” to help creditors decrease risk and reduce losses while approving more profitable applicants.
What Are the Options?
The main options for scoring models boil down to this:
- Custom (too expensive) – Requires archive data (data from an historical point in time) that smaller banks and credit unions don’t have, and it can take many months to develop scores and implement
- Expert (too subjective) – Expert models are created by drawing upon models we’ve developed for clients in similar regions, decision areas, and product types, which are combined with your knowledge of your portfolio and unique business environment
- Pooled (just right) – Ideal for customers who want greater predictability than a Generic or Expert score, but don’t have the historical internal data required for a custom model. No data is required from the customer because it’s built on pools of historical data from other financial institutions.
What Are the Benefits of Using a Pooled Model?
Here’s what an organization can expect when it comes to benefits after adding a pooled model to the origination process:
- A pooled model helps your organization make more precise, value-based decisions during the originations phase.
- It increases productivity significantly because it automates decision-making. This decreases the time needed for credit analysts to review and investigate each loan.
- It enables you to approve more applications and increase your profitability through enhanced client experience.
- Using a pooled model reduces your delinquency and charge-off losses.
- It quantifies the risk of your portfolio – this is important for management, so you can monitor and adjust the amount of risk to take on based on overall appetite for risk and external factors.
- It ensures regulatory compliance by generating intuitive reason codes, so you can respond to legal concerns and customer questions.
- A pooled model also provides objective, consistent decisions. With scoring, there is a specific cut-off score that credit analysts use to accept, reject, or refer the application for further investigation. This ensures that there is consistency in decisions across all credit analysts, and the easier cases are automatically decided without analyst intervention.
Why Aren’t More Organizations Using Pooled Models?
Given the many benefits, why aren’t more enterprises using pooled models when making originations decisions?
Most enterprises (unless you’re a Fortune 1000 company) don’t have the time, money, or resources to deploy and support complex custom models. It’s all extremely complicated and difficult to manage. On top of that, there are regulatory concerns, as well as concerns about explainability and disruption to operations.
The ideal target audience for pooled models are small and regional banks and credit unions that currently use either: a generic score, and/or judgmental score or rules for their originations decisioning. Creditors who want to move up the analytics scale to something more predictable should seriously consider using a pooled model.
Pooled models can be developed and deployed quickly, which makes them a sweet spot for lenders who don’t have enough data to create custom models but are seeking the flexibility to grow their portfolios responsibly.
Pooled models deliver consistent scaling across products to allow lenders to compare risks between different products. Industry-leading models, like FICO® ARM 4.0, use scorecard technology enhanced with machine learning (ML) techniques and an innovative and robust bureau characteristic library to capture even more facets of information.
In our next blog, we’ll discuss the incremental performance (predictability) you can expect when using a pooled model in your originations decisioning.
Contact FICO at email@example.com to learn more.