The Reserve Bank of India on Monday introduced draft regulations aimed at enhancing the governance, development, and validation of credit risk models used by regulated entities (REs).
Recognising the potential risks associated with model-driven decisions, the central bank seeks to establish a robust framework for managing these complexities.
According to the RBI, models are integral to the credit management lifecycle, influencing borrower selection, credit scoring, pricing, risk assessment, and provisioning.
However, their reliance on assumptions can expose REs to model risk, impacting prudential, compliance, and reputational aspects.
“While the application of technology in models has facilitated faster decision-making under complex scenarios, it also adds complexity to the model risk management framework implying the need for a comprehensive understanding, a robust validation mechanism as well as appropriate governance and oversight,” the RBI said in its release.
To mitigate these risks, the draft regulations mandate a comprehensive Board-approved policy for all deployed models. This policy should encompass governance, oversight, development processes, documentation, independent validation, change control, and monitoring. The RBI also emphasises the importance of a model inventory and a clear approach to using third-party models.
The deployment of individual credit models requires the approval of the Risk Management Committee or a designated sub-committee. The central bank outlines broad principles for model development, including clear objectives, robust inputs, detailed documentation, scalability, and explainability. Both internally developed and outsourced models must adhere to these standards.
To ensure model reliability, the apex bank mandates a vetting and validation process, independent of model development.
Models should undergo validation before deployment and periodically thereafter and the banking regulator reserves the right to conduct its own model validation or engage external experts.
RBI defines credit risk models to be any quantitative method that applies statistical, economic, financial, or mathematical principles and assumptions to process data into an output to be used for credit decisions.
Stakeholders have until September 4, 2024, to provide feedback on the draft regulations, which will come into effect within three months of issuance.
