Credit decisioning — how to go from confusion to conclusion
Why 'fast and frugal' reasoning combined with the computing power of 'business rules engines' can be gamechanging for digital lending in India
Across the annals of human history, decision-making has intrigued thinkers, philosophers, and scientists alike. We have come a long way — from seeking guidance from the stars to being trapped in the iron cage of rationality (rigid logical thinking) to finally gravitating towards a semblance of balance that is human and rational. Over the years, we have steadily been coming to terms with constraints — both contextual and psychological — on our ability to make optimal choices.
Having humbly submitted to the fallibility of decision-making, I believe we should strive to achieve, if not optimal outcomes, at least acceptable ones, especially at the organisational level.
Much has been said about administrative decision-making, but the German psychologist Gigerenzer’s take on decisioning struck a chord with me. He urges us to make a virtue of our limited time and knowledge by mastering simple heuristics, an approach he calls “fast and frugal” reasoning. I say, add computational might to it, and enhance this method of fast & frugal reasoning, something we have been trying to do at FinBox to solve for the myriad challenges in digital lending.
“Truth is ever to be found in simplicity, and not in the multiplicity and confusion of things”
- Isaac Newton
Risk is an inescapable part of every decision. To make even ‘acceptable decisions’, lending organisations must be able to calculate and manage the attendant risks — the possibility of loan defaults, financial losses, regulatory issues, and other adverse outcomes that lenders must account for as part of credit decisioning. These so-called attendant risks are now to be managed within the constraints of constantly evolving consumer expectations.
The decisioning dilemma for digital lenders
Digital lending players are primarily concerned with two questions: One, how to ensure that customers are who they claim to be without putting them through cumbersome journeys? Two, how to build lending policies that approve more customers without raising the limits of risk?
Now these two problems are more intricate than they initially appear. The first problem i.e, the identity verification process must navigate the dual objectives of KYC/AML compliance and fraud checks, while minimising friction in customer experience.
The second challenge pertains to creation of underwriting policies that foster inclusivity and drive business growth without elevating risk or compromising on customer convenience.
The embedded finance revolution has complicated matters further, as these goals are now to be achieved within digital lending journeys involving multiple, complex partnerships. With increased outsourcing and third-party participation in lending customer journeys, risk has increased manifold for lenders.
Add to this the challenge of having to contend with constantly evolving regulations, compliance requirements, fraud landscape, macroeconomic conditions, and various other factors.
FinBox Sentinel - Helping lenders automate credit decisions at scale and speed
After years of close collaboration with lenders, we have developed a deep comprehension of the complexity of the challenges they face. This is exactly why we built Sentinel — a low-code business rules engine (BRE) powered by big data analytics.
In essence, it provides a robust interface that empowers business experts or risk teams within lending organisations to automate lending processes efficiently, leveraging targeted insights, all while ensuring an exceptional customer experience.
To automate credit-decisioning you don’t need Laplace’s demon-like softwares that integrate a zillion data sources. You need a BRE that connects you to multiple relevant data sources and also allows you to optimise for it. For instance, if the decisioning outcomes for a specific user base using bureau data alone and bureau + bank + GST are no different, why use three in the place of one, only to add cost.
It would therefore make a world of difference if a BRE supported all relevant data sources to begin with and tells you over time, why you needed only one, two, all three, or more. This is what I mean by frugal. Now let me give you a sense of what I mean by fast.
Risk teams at lending organisations understand the business best. However, they feel disempowered when it comes to taking action because of their dependency on tech teams to execute policy changes. If business rules are embedded in code, even the slightest modification is bound to take four weeks.
The need of the hour is a solution that simplifies credit decision automation — a single interface that lets you combine multiple data sources into a single workstream, test it then and there, and deploy it at the source of loan origination within a matter of minutes.
Equipping lenders to serve underbanked segments, without stretching risk limits
A coin has two sides, and one side cannot exist without the other. Similarly financial inclusion-driven growth cannot exist unless lenders are equipped to meaningfully assess creditworthiness of unbanked and underbanked segments.
First, they should be enabled access to a diverse set of powerful alternative data sources in real time and a rich repository of historical data of new-to-credit users. Add to it the power of demographic insights at data source level, feedback loops, simulation capabilities, and the ability to test new strategies in a live production environment. Given these technological advancements in a BRE, it would equate to the distinction between a workhorse for labour and a thoroughbred for combat.
Empowering lenders to combat fraud and meet compliance requirements from day one
The fraud landscape is constantly evolving. It was reported that escalation of digital fraud in India has led to businesses investing USD 7.6 billion to stop financially-motivated fraudsters alone. Organisations lose an average of £3.5 million in revenue due to a single non-compliance event. However, surprisingly 40% of businesses use office productivity software, such as documents and spreadsheets, for compliance.
A BRE that aggregates risk and equips lenders with real-time analytics while empowering them to implement risk responses swiftly, would be a timely solution. This explains why 88% of IT decision makers are considering modernising their business rules engines as per a survey conducted by Mckinsey involving 261 businesses.
A BRE should be able to seamlessly combine fraud, KYC/KYB into one workflow with drag-and-drop simplicity. The process of customer verification becomes significantly streamlined when equipped with the most relevant and comprehensive data. By simultaneously accessing multiple data sources, it becomes possible to validate information as it flows through various stages of the workflow. As a result, the need for unnecessary document requests is minimised, allowing for faster and more efficient customer verification.
The bottom line
This is the time for digital lenders in India to gear towards a breed of technology that improves strategic decision-making across all tiers of an organisation, for we stand on the brink of a Cambrian explosion in digital credit. As ONDC-like frameworks espouse embedded finance, credit will grow by leaps and bounds. However, the decisioning that backs it is hamstrung by inflexibility — which can be likened to what Max Weber referred to as the 'iron cage of rationality' on a sociological level. I eagerly await your thoughts on the matter.
If you’d like to know more about Sentinel and our other products and platforms, please don’t hesitate to reach out to me personally at rajat@finbox.in !
There is no doubt that digital lending space has done wonderful job. I being in banking for last 18 years & shifted to loans sourcing in my own venture of helping customers( Rejected by banks) to get credit. To my surprise within these limited period of last nine months i am very much surprised to see how use of CIBIL data is helping & at the same time denying credit to many financially needy ( but imprudent customers - not there fault) . They have clear case of loan to be disbursed but due to poor CIBIL they are straight away no. There is urgent need to develop another credit eligibility matrix consisting of various other parameters which are getting ignored like Reason of default( not just days of default) - specially in case of business's having track record & promoters experience, Nature of job, Family data & there are "n" parameters which are getting ignored & credit being denied & then they have to go to private money lenders to get loan at 3 - 9% monthly pay. This also brings to point that there are enough opportunities for different set of lenders to create huge business in lending space which looks risky due to not updated risk measurement tools/ local know how or customer level knowledge & taking all customers as same due to CIBIL set standard. This is allowing duplication of credit to those who are already have "n" offers & denying to those who needs credit. Still there is lot of work to be done, profit to be earned & capital to be deployed or risk capital to serve the need of vast set of customers. Required is more & newer set of risk management tools/ practice's & people. Use of psychometric tools which can predict better outcome by capturing multiple data sets not just CIBIL score. With so much advancement in DATA mining this should not be issue.