Business Objective
Our client is a leading mid-sized regional bank in the US with a diversified consumer lending portfolio.
Our objective was to build a robust loan-level loss forecasting model for their credit card portfolio, along with detailed documentation of the process that would successfully meet the guidance and expectations on Current Expected Credit Loss (CECL).
Challenges
- Constantly evolving guidelines – FASB and the banking regulatory bodies such as FRB, OCC or FDIC has not specified loss calculation methodology
- Substantial data preparation – the need to summarize defaults, credit and fraud losses, and to create monthly trend information
- In-depth analysis of economic risk factors at a granular level – the macroeconomic forecasts could not simply be based on deviations in the past.
- Extensive documentation needed to keep the model audit-ready
Solution MethodologyÂ
- Leveraged both bottom-up (loan-level) and top-down (segment-level) approaches
- Created segmentation based on delinquency, payment activity, and tenure at observation points separately for each of the dependent events (credit charge-off, bankruptcy and prepay)
- Developed a comprehensive loss forecasting framework with justifiable assumptions for calculating PD (Probability of Default), EAD (Exposure at Default) and LGD (Loss Given Default)
- Developed an ensemble of models alongside business context to forecast potential losses over the lifetime of the loans
- Back-tested the models along with testing for different stress scenarios
- Worked with the bank’s finance group to simulate ACL calculations and comparisons on incurred loss methodology and CECL methodology
- Established detailed and comprehensive documentation that described the process end-to-end and also included quantified justifications for all the decisions and assumptions
Business Impact
- Ensured the client had a robust loss forecasting model that met regulatory guidance and expectations on CECL
