The benefits of custom scoring models are becoming more obvious and the cost has plummeted, but little attention is being paid to developing better ways for risk managers to implement them.
Resource constraints, cost justification and extended timelines are usually the biggest hurdles. Risk managers need to demand strategic scoring platforms that meet their institution’s credit scoring needs. Systems that easily accommodate custom scorecard use help risk managers focus on mitigating risk and improving the value of credit decisions.
A well-rounded, strategic scoring platform should use custom scoring models, generic credit bureau scores and data from other sources – including credit reporting agencies. Reasons for the widespread use of generic credit bureau scores, such as the FICO score, is that these scores provide easy implementation and are available from all three major credit bureaus – Experian, TransUnion and Equifax. These scores can be incorporated into origination, prescreen and account-management strategies.
For institutions with enough internal performance data that are willing to make the financial investment, custom credit scoring models provide an added competitive advantage because of improved predictiveness.
Ideally, an institution’s credit scoring system should provide all of the benefits of generic bureau scores, but this is not always the case. A custom scoring project traditionally can take anywhere from nine to 15 months, from a vendor search to full integration testing on the loan origination system.
Another drawback is the tendency to treat each generation of custom models separately, starting from scratch each time. A strategic view of any organization’s custom scoring requirements will streamline scorecard implementation, giving greater benefits and improved competitive advantages in lending.
The Right Choice
The first strategic decision involves choosing a set of credit bureau attributes, the building blocks for credit scoring models used within application decisioning strategies.
The chosen set should include a library of credit attributes relevant across the credit bureau’s data and the institution’s lending products. This set should be updated as credit reporting rules and consumer behaviors change. Because few financial institutions have such expertise or resources, a better option is to partner with a third party that has the ability to create custom attributes and enable its clients to modify them if desired.
Regions Bank, based in Birmingham, Ala., partnered with Digital Matrix Systems Inc. in Addison, Texas, to gain access to the DMS Summary Attributes, a set of 2,400 standardized attributes available across Experian, TransUnion and Equifax, the two Canadian bureaus, and each merged combination of those credit files.
A select set of these and custom attributes are used by the bank in credit scoring models, policy enforcement and other automated decisioning criteria. Because these standard attributes are developed and maintained by a third party, scorecard attribute testing by the bank is practically eliminated, enabling new custom scoring models to be put into place in just one or two months.
The second strategic decision is to determine how to access the third-party attribute library through the financial institution’s online application processing system. Clearly, a financial institution prefers a service requiring minimal impact to its application processing system and minimal internal IT resources.
The service needs to be flexible so that few changes are required when the next set of custom models, using perhaps entirely different attributes, are developed. Because institutions often are tasked with implementing scores across multiple platforms, the attribute set needs to be easily accessed by a variety of systems.
Having evolved through mergers and acquisitions, many legacy processing systems are inflexible or resource-intensive when considering additional data interfaces. This complexity results in time-to-market costs that are consistently underestimated. Implementation platforms are available that interface with legacy processing systems in their traditional methods and formats while opening external data sources with speed and flexibility.
To resolve these issues, DMS integrated its platform with Regions’ legacy application-processing system. The platform uses credit bureau native inquiries and responses to interface between credit bureaus and legacy systems. Integration is as simple as redirecting credit bureau traffic to the platform and interpreting any new score or variables returned to the origination systems. This enables the bank to react to the changing credit landscape and to quickly modify its custom scoring models.
A strategic scoring platform essentially eliminates the implementation hurdles associated with traditional custom scoring model projects. It also succeeds in bringing the focus back to solving the many risk-management and competitive challenges faced by banks in today’s economic environment, rather than focusing too much on implementation obstacles.
Tom Bloetscher is senior vice president of consumer credit at Regions Bank. He can be reached at email@example.com. David Graves is vice president of decision science at Digital Matrix Systems Inc. He can be reached at firstname.lastname@example.org.
To comment on this article, contact Darren Waggoner, chief editor at Collections & Credit Risk, at email@example.com or 815-463-9008. Your feedback will help shape CCR's ongoing coverage.