The amount of Expected Credit Loss (ECL) necessary to support a credit portfolio depends on the probability distribution of the portfolio loss. By using the aggregate credit quality as a proxy for the ECL, the portfolio is subjected to risk arising from the loss of granularity. All things being equal, ECL when applied at the top-down level, offers important portfolio insights, however, does not adequately differentiate the credit quality of the borrowers and discrete loans.

Traditional underwriting framework focuses more on collateralized lending and less on the qualitative and quantitative factors that form the basis of a credit score. Credit scores are organized in buckets, with each bucket mapped to the probability of default (PD) by calibrating the credit score. This is a complex quantitative operation and requires expert level input.

For most lenders, over-collateralization remains a sufficient condition to grant the loan. Whilst this might prevent the incurred loss, i.e., expected losses attributed to loans classified as ‘Stage 3’ under IFRS 9, for loans in Stages 1 and 2, it is important to understand the probability distribution of the default rates at a more granular level. This can be achieved by developing a granular rating scale. Imagine if the portfolio is distributed across wider rating buckets, the overall impact would be lower PDs within each bucket and hence lower ECL.

Given ECL is a function of Probability of Default and Loss Given Default (LGD), lower PD assignment implies lower ECL. With more granular rating buckets, relatively lower concentration of loans in each bucket can also be achieved and hence a lower skew in the probability distribution. Advanced practice regulators require that any one risk rating must not represent more than 30% of its credit portfolio.

Lenders also seek to increase granularity for the purpose of risk-based pricing, since the more granular a lender’s rating scale, the less opportunity their competitors have to cherry-pick with more granular rating system.

Another reason to revamp rating schemes to increase visibility through granularity approaches is that, historically, borrower rating and facility rating have not been separated. That is, where secured loans to weaker quality borrowers are recorded at higher ratings because the collateral alone, reduces the financial exposure.

Simply put, granular rating scales enable the ability to zoom-in on rating hot spots, and strategize accordingly, including to develop mitigation measures that can be adopted pre and post-loan origination. Such diligence enables calling into question the predictive power of credit scores through back-testing or observed default rates.

At BankingBook Analytics, our ECL application uses advanced dashboarding to empower executives and management visualize the overall view and distribution of loans in Stage 1, 2 and 3, using Google Maps API. This wide lens depth and breadth visibility enables the development and implementation of mitigation strategies together with lending strategy and loan origination teams.

Best practice banks invest heavily in their analytics tools to use a 25-bucket rating scale. BBA is an expert at exactly that, without the big bank budgets. We assist banks, credit unions and lenders of all sizes to develop and implement granularity in their rating scales to improve their credit scorecards – and most importantly – to assess and maintain strength in their portfolios. Our tools and approach help to prevent and reduce loan losses by having a positive impact on ECL and profitability.