Overview

Given the announced changes in regulatory requirements, intensified supervisory scrutiny of models in light of the on-going COVID-led challenges, the time is ripe to consider model re-development or re-calibration. Our paper makes the case for a default pooling solution to solve the ‘default scarcity’ challenges in the low default portfolios (LDP).

Key Take-aways:

• It is widely acknowledged that credit modelling teams at banks face significant challenges in developing rating and probability of default (PD) models for LDPs because of sparse defaults.
• With global regulatory changes (Basel III reform) and heightened supervisory scrutiny (TRIM remediation), the time is ripe to consider re-development of LDP models. Banks need to be careful in electing a model methodology that meets the multiple requirements of regulatory compliance, robust model performance and optimal regulatory capital charges.
• Default data pooling is a compelling proposition for PD and rating models of LDPs, especially for the ‘banks’ and ‘large corporates’ portfolios. These portfolios have a high likelihood of arriving at a rich data pool through a consortium construct.
• The option of leveraging a data pool with obligor-level data and binary classification of obligors as ‘defaulters’ or ‘performing’ would offer banks to meet the core requirements of portfolio representativeness and transparency, as compared with alternative options of shadow ratings or PD pooling consortiums, where the output is subjective and arguably, a ‘black box’ estimate of PD.