Lenders have sometimes resisted automated decisioning for fear of losing the internal credit scoring expertise they have spent years developing. A new upgrade to ADP means that this is no longer a cause for concern.
For many lenders, their back office operations are a source of some pride.
When they look at their underwriting teams, they often see a group of professionals who have become experts at screening for applications that meet their precise appetite for risk. These individuals will be using in-house scorecards that have been honed over time to achieve what they think is something close to perfection.
The scorecards are the product of the lender’s accumulated knowledge, built-up over sometimes many years of lending. They are expertly adapted to market changes and specific loan book requirements.
Little wonder that when some lenders contemplate what they see as a ‘big bang’ move over to automated or intelligent underwriting, they approach it with some degree of trepidation. The transition can seem like too big a leap for them.
They have feared that their hard-won wealth of knowledge and expertise will be overlooked and, in its place, they will have a hyper-efficient regime, but one that is non-specific to them and their needs.
This has often been at the back of the minds of some of the lenders I have spoken to since LendingMetrics began implementing Open Banking and automated decisioning platforms five years ago. This has been an understandable perception as legacy scorecards are often used to guide the implementation of automated decisioning platforms.
Up until now, there has always been the risk that something may be lost in the digital transition, given that legacy systems are built on much older foundations.
This is particularly the case because they often use program languages such as Python (V1 and V2), R, SAS, PMML and JavaScript that do not converse easily with automated platforms. So tapping into scorecards’ undoubted value was therefore not always straightforward and usually involved employing a number of work arounds.
However, as of this month, this has all changed.
Following a 14-month development program, LendingMetrics’ development team have introduced a new version of ADP (Auto Decision Platform) that utilises a universal risk model adaptor, allowing the inclusion of multiple external languages. This development – unique to LendingMetrics – means that, for the first time, legacy systems can fully interact with all of ADP’s functionality, and ADP can fully interact with them in return.
Credit risk teams long accustomed to the capabilities of Python, SAS, R, PMML and JavaScript, will be delighted to learn that all are supported by this feature and, due to its “native inclusion”, the risk models can seamlessly interact with ADP’s full range of features. These include the engine editor, which allows users to create, edit and deploy decision engines in real time, and the warehouse, an optional module designed to assist with long-term storage of decisions and decision data.
It goes without saying that this upgrade makes the job of the credit risk team much, much simpler. From now on, all they need do is manage one single platform and focus on increasing conversions without increasing costs.
At a time of economic uncertainty, when accurately assessing risk has never been more important to lenders, seamlessly combining years of lending wisdom with automated decisioning has never been easier, or such a good idea.
David Wylie is commercial director of LendingMetrics
