Rethinking the Automation of Loan Underwriting
For years, banks have been automating inefficient steps in loan underwriting. Over time, more aggressive institutions have been able to automate the majority of the process—some have even been able to fully automate account opening. Like the last nine in six-sigma, the final steps of automation have proven elusive. But, there is light at the end of the tunnel as new technologies open the door for straight-through-processing of loan applications.
The Benefits of Automation
Historically banks have implemented most automation projects with the goal of improving upon existing manual processes. Two advantages to this approach are 1) IT development costs are easily justified by improvements in productivity and 2) the risk of failure is minimized. The downside is the institution will never be able to move past the inherent limitations of their existing manual processes. Preceding the crash in 2007, several institutions used automation to improve efficiency. Without vetting their lending policy before automating, they accelerated the number of bad loans approved. Their ex-employees can attest to the dangers of not automating carefully.
An effective account opening process needs to start with the assumption that all applications should be automated. Rather than using an exception-based approach that focuses on manual review of any problem, the system should incorporate exception handling into the normal flow. Straight-through- processing techniques used in trading have applied this concept for some time and proved its efficacy.
Facing the Challenges
The first challenge in implementing fully automated account opening is to build a methodology for handling mandated exceptions and stipulations. For example, an address mismatch from the credit bureau or a fraud indicator requires a manual review because you cannot decline credit for a non-credit reason. Fortunately, there are a number of data resources available today that can handle every exception we’ve encountered. Even a questionable identity can be confidently evaluated using additional data sources.
The next common challenge we see is that applications removed from the automated pathway are rarely able to return to the underwriting process where they left off. For most institutions, this means that any manual review requires the application to be reentered or at least reprocessed. Aside from the obvious data entry errors that are common with this approach, the user experience and timeliness of the application are destroyed. As an alternative, for more than 15 years Zoot has provided the ability for applications to enter a manual review queue, be cleared of any stipulations, and then reenter the automated process at the precise point they stepped out. While our goal has always been to eliminate all manual reviews, it is critical that any lingering exceptions be handled elegantly.
Processing applications in near realtime provides an opportunity to keep the consumer in the underwriting stream. This means that any potential data entry error or unexpected credit issue can be addressed directly by the consumer as part of their application process. This eliminates the need for costly follow-up and application delays. The user experience is also improved when the consumer senses an interactive relationship as opposed to a static web form.
Automation in Action
So what does this look like in practice? Obviously, the consumer will start with an optimized application interaction that validates all data as it is entered. In electronic channels, the user’s identity and device’s reputation can be analyzed as well. If the application comes through a mobile device, image capture can be used to obtain documents and the user’s likeness. Once submitted, the application begins the underwriting process. Data acquisition and decisioning can typically be completed in under a second. Any exceptions are routed to automated queues for instant review.
Here are a couple of examples of how exceptions can be handled through data and service providers. An address mismatch can be cleared using LexisNexis Instant Verify. A thin file or no hit from the credit bureau can be addressed using alternative data such ID Analytics Credit Optics to provide a credit decision. For stipulations that can’t be cleared automatically using additional data, the user interface should be designed to support additional questions. The manual review is then completed by the consumer in realtime. Even with multiple stipulations, the entire process (except additional user interactions) should complete in just a few seconds.
Capturing the Elusive
We are on the cusp of a major change in the way banks process loan applications and consumers expectation of a hassle-free experience. Embracing a straight-through approach today will help institutions rethink their processes based on solid risk management needs, rather than long-established practices from a bygone era. We are seeing this approach applied in the mobile space already, but are just beginning to see early adoption of straight-through processing across all channels and all product lines. With a disciplined methodology, lenders can finally automate the last elusive steps in their underwriting process.