Loan default prediction and collections aren’t always showcased at the leading edge of digital transformation. Nevertheless, efficiency in these processes for lenders (or their service providers) has a direct impact on the bottom line. These capabilities wait to be tested when an economy softens.
Automation of case management, legal processes, credit agency interaction and similar has typically been where most providers have invested technology effort. Solutions are often localized to align with local lending and legal processes.
Basic segmentation is also sometimes performed semi-manually — particularly for small business lending, where industries that experience an easily identifiable downturn (for example, agriculture) may be forecast to experience higher default rates.
In some markets there has been creative sharing of contact and other information to help better trace debtors and understand behaviour.
What is emerging, however, is the ability to leverage the same data underpinning other customer insights for upsell, cross-sell, fraud detection, customer segmentation and initial credit decision making in the default prediction and collection processes.
Lenders may, for example, segment customers based on default risk (models for which may be very closely aligned with credit decision models) but there is also an opportunity to create segments to align with recovery approaches that may yield the best chance of success.
Real-time behaviour information — useful for identifying unusual behaviour patterns that may indicate fraud or account takeover, or present opportunities to present an offer — may also be useful for collections. For example, to encourage customers to make a repayment at the most opportune time based on their cash flow. However, any such effort needs to tread the careful line that other customer-focused banking analytics must follow — ensuring purposeful and compliant use of data for customer benefit.
Consider the specific opportunities and technology that may be applied to take advantage of these opportunities.
- Predict borrower difficulty early — with models that work across wider sets of data than just pure repayment schedule adherence — for example, incorporating models built using transactional data that may be available to the lender. Pure play lenders are in a less strong position to do this unless they are partnered with others with available data and insights.
- Leverage a 360 degree view to determine borrower options with improved chances of success. Such options might include consolidation of loans, consolidation or adjustment of insurance related to loans, repurchase options on other asset backed-finance options and more.
- Incorporate rescheduling assistance offers, repayment deals and more into real-time ‘Next Best Action’ type technology.
- Leverage machine-learning based segmentation to establish risk groups, optimal communication channels and repayment offer structures.
- Experiment with digital portals for customers to model rescheduling and other options — even prior to default.
Some of the technologies that may support the above include:
· Data hubs — possibly incorporating real-time transactional data
· Machine learning models applied for both making early default predictions and clustering
· APIs to expose data internally for actions / interventions to be presented through inbound and outbound communication channels
· APIs (and associated security solutions) to present data or actions to outsourced collection agencies (if used and to the extent deemed permissible)
· Digital portals and correspondence solutions.
For bank lenders who execute collections in the first instance themselves, the opportunity for digital transformation of the process is significant and should be seized before a lending book is tested in times of economic stress.