Early Detection of High-Risk Customers Helps Avoid USD 1.5MM in Repossession Losses

Business Objective

Our client, part of a multibillion-dollar automotive company, is a leading India-based provider of durable consumer loans specializing in motor loans.

The management wanted to optimize the entire customer financing journey with the end goal of reducing vehicle repossession losses of USD 3 MM annually. The objective was to identify the customers (potential and existing customers) who will be prone to repossession and estimate the loss due to repossession.

Challenges
  • Manual loan management process, leading to inconsistencies across the data
  • Missing data for certain key attributes that would help estimate the losses
  • Flat data files required sequential processing that increased the processing time

Solution Methodology
  • Developed three different predictive models – one for prospects at risk of repossession, one for existing customers at risk for default, and one loss estimation model
  • Built the Acquisition Risk Scoring model for determining the probability of default for potential customers
  • Developed the Repossession Risk Scoring model, wherein existing customers were categorized into five default risk bands (very low to very high). Transformed collection data to understand payment behavior and determine the probability of default
  • For Repossession Loss Estimation, first predicted vehicle resale value using ML techniques and then computed the losses using other factors like Installment Overdue Amount, Future Principle, Resale Value & Advanced EMI
Business Impact
  • Avoided USD 1.5 MM in repossession losses due to early detection of high-risk customers at the application stage

  • The model accurately identified 85 % of all repossession cases during validation

  • Developed a what-if scenario tool for loss estimation

Copyright © 2023 Tiger Analytics | All Rights Reserved