Credit Risk Modelling
Introduction
Credit risk modelling is more and more at the heart of the banking business as it intends to allow banks to effectively measure the exposure to credit risk, the most prominent risk banks are typically exposed to. This measurement can be done (i) for regulatory purposes, when internal models (under the so-called IRB approach) play a fundamental role in the calculation of risk weighted assets (RWA) and/or for economic capital computation purposes (Pillar 2), (ii) for the calculation of impairments under IFRS 9 (international accounting standards ruling impairment determination), or (iii) for multiple business-oriented decisions (for credit granting, monitoring or recovery).
As an area subject to increased supervisory attention and highly regulated since mid of last decade, technological innovation is also playing its role in increasing model performance, namely with the emergence of Generative AI (GenAI), the use of machine learning techniques and, most recently, where climate-related and environmental (C&E) risk factors have to be integrated in credit risk.
Challenges you might be facing
- Simplification of IRB model landscape
- Integration of C&E modelling into the IRB space
- IT implementation of new models
- Interconnection between IRB and IFRS9 modelling frameworks
- New generation of models for decision purposes (ML/AI)
Simplification of IRB model landscape
In anticipation of the adoption of CRR3 (Basel IV) and as reaction of the recurrent ECB’s appeal for model landscape simplification, several banks are revising their plans for the future IRB perimeter, which implies a revised combination of reversions to SA, changes of scope of the PPU usage and, for certain portfolios, a step back to F-IRB. All this shall take place in a context where the ECB shall be very keen in understanding the criteria for reversion to SA/F-IRB approaches (e.g., operational effort and costs, data availability, impact on capital requirements, impact on risk management) and shall expect banks to submit a single comprehensive and consistent request for all the rating systems.
Banks shall so need to perform cost-benefit and feasibility assessment across all rating systems, submit a formal request to the ECB and for the segments that potentially shall not remain under the IRB approach, define how other models will be affected (e.g., IFRS9, Pillar 2).
Integration of C&E modelling into the IRB space
The growing expectations in terms of integrating C&E considerations in the credit lending process have been requiring modelers to start testing additional risk drivers related to C&E, when modelling IRB risk parameters, incorporating those that are found to be relevant and material. The challenges are still significant as limited historical data on ESG do not ease the task of meeting the so demanding IRB modelling requirements for the different risk parameters.
Nevertheless, banks shall need to (i) accelerate the processes for identifying, collecting and effectively manage relevant C&E data, (ii) include C&E risk drivers in a long list for future IRB model redevelopments, and (iii) review override framework for the immediate possible incorporation of C&E risks.
IT implementation of new models
The revised version of the ECB guide to internal models, published in February 2024, clearly set the ambition of banks being capable of implementing material changes or extensions in a timely manner upon receiving permission, i.e., in a timeframe no longer than three months from the date of the notification.
All this requires banks to significantly re-think their current capacity to ensure that data for modelling and implementation are sourced from single “ready for use” environments, allowing fast tracks from modelling to implementation and supporting parallel runs of non-approved models.
Interconnection between IRB and IFRS9 modelling frameworks
A growing trend in the market is leading to a search for the maximization of synergies across IFRS9 and IRB frameworks: data collection processes, risk inputs, segmentation drivers, testing batteries are examples of areas where both worlds touch each other, aligning planning and standards.
The same holds true for enhanced stress testing capabilities in a context where macroeconomic uncertainty and growing ESG concerns claim for increased realism and coherence across models (IRB and IFRS9) and assumptions.
New generation of models for decision purposes (ML/AI)
Banks are now accelerating their efforts in applying machine learning and artificial intelligence (ML/A)I techniques for the prediction of key business outcomes based on past experience at the various steps of the credit process, namely underwriting (credit analysis and decision), credit portfolio monitoring (including EWS and reporting) and recovery. There are already a few successful experiences in introducing ML within IRB models, so one may expect soon further developments in the market.
How can we help
PwC has supported numerous banks, across several geographies, in addressing their credit risk modelling challenges, particularly in the IRB space. In fact, PwC set up a dedicated team in 2015 – year where EBA announced its first plans for revising the IRB framework – that since then has not only closely monitored all key regulatory and supervisory developments, but also been highly engaged in supporting clients throughout the TRIM exercise, in the implementation of IRB repair programs, in the simplification of the IRB landscape or already in the adoption of Basel IV IRB requirements.
As such, we are in great position to assist you with the following topics:
- Carry out cost-benefit and feasibility assessments across rating systems to support all needed decisions in relation to the future IRB perimeter, inclusive in the preparation of the formal request to be submitted to the ECB.
- Support in the integration of C&R risk factors within the IRB framework, starting with the identification of relevant C&E risk factors and concluding with the respective incorporation in the final model design, calibration approach and/or override framework.
- Help in re-thinking the target model deployment capacity so as to allow for a timely model implementation.
- Enhance synergies across the IRB and IFRS9 frameworks converging in what is common and justifying the divergent assumptions and methodological choices.
- Support clients in understanding the potential usage of AI/ML in enhancing the performance of their credit risk models, having in consideration the best market practices.
- Provide strategic, technical and project management expertise for the implementation of CRR III as well as in IMIs (internal model investigation) mission.
