The Covid-19 pandemic has created high uncertainties on a global scale and institutions face unprecedented challenges in accurately representing their portfolio risks to the market when the crisis has not yet fully materialised. The guidance published by regulatory agencies and industry groups are not exhaustive and places more weight on judgement rather than mechanistic triggers thus driving further divergences in the credit measurement approach. It is therefore crucial that institutions have robust procedures to appropriately quantify their credit risks that is flexible enough to allow management intervention, timely to reflect all guidance and relief measures, efficient to allow quick turnaround of results (including sensitivities) and robust so that further enhancements can be incorporated thus serving a long-term need.
This paper presents Mazars’ Expected Credit Loss (C/ECL i.e. CECL and ECL) tool that is an end-to-end solution to assist institutions comply with the IASB (IFRS 9) and FASB (ASC Topic 326) accounting standards as well as incorporate Covid-19 measures throughout the modelling process.
The Covid-19 global pandemic has far-reaching consequences and institutions will need to adapt their provision modelling approaches to ensure they can best reflect government relief measures and regulatory guidance within their processes. The quantification of risk is a critical step in any risk management framework and having flexible yet robust tools will help institutions position themselves well through current and future challenges they face.
The provisioning tools are not the same across institutions because of numerous factors such as organization size, available resources, urgency to implement (including expectations from regulators) and in-house technical expertise. As such, rigid processes and ‘black-box’ provisioning tools with sparse documentation is not uncommon. With Covid-19, these institutions face high challenges to incorporate relief measures and judgements. Key risks that institutions should address, related to Covid-19 and long-term provisioning requirements, are as follows.
- Operational risks on data entry, formula references and complicated update procedures
- Scope narrow as focused on provision calculation and not input verification
- Time-consuming to perform an end-to-end calculation
- Rigid procedures resulting in limited / no manual judgements
- Black-box solutions making it difficult to audit and understand
- Documentation gaps making it difficult to audit and understand
- Costly to maintain and update as require external support
- Overlays exogenous to the model thus making it difficult to interpret disclosures
Mazars has significant experience in the development and implementation of expected credit loss models for the IFRS 9 and ASC 326 accounting standards. Our team combines the expertise of accounting and risk management professionals with deep industry expertise who understand the specific requirements of regulatory standards and their impact on your organization. The methodologies are tailored to your credit risk dynamics and Mazars suite of tools can perform the end-to-end calculation of provisions covering data analysis, calibrations, provision estimation and reporting disclosures.
Since the onset of Covid-19, Mazars has enhanced our tool offerings as follows:
Management ECL Overlays
- Non-modellable risks can be incorporated into provisions
- Overlay segment and allocation method
Management Stage Override
- Manual adjustment to stage (IFRS 9) or default trigger (CECL)
- Audit trail of stage/default overrides
Additional Downturn Scenarios
- Up to five scenarios to incorporate more information in provisions
- Weights and forecasts input by user / external source
PD Scenario Reversion
- Speed of mean-reversion by scenario
EAD Adjustment for Payment Moratoria
- Projected balances reflect moratoria/payment holidays
- Revised EAD is an input to the LGD
Disclosure Attribution for Overlays
- Separately show impact from overlay amount at each reporting date
For more information on how we can help, please contact: