The Reserve Bank of India (‘RBI’) recently issued a draft guideline[i] open for public comments for regulated entities (‘REs’) to develop and validate their credit risk models. The primary objective is establishing a robust system for tackling credit-related problems and minimising capital loss. The new regulatory principles of management of credit risk model are set to come into effect within three months from the date of issuance, and the guidance note on credit risk management (ch. 3 – credit risk models), dated 12.10.2002, has been repealed.
Key Provisions
The REs are instructed to formulate a detailed policy, which their respective boards must approve. The policy must cover all substantial aspects of the model, including its functioning, governance, procedures and compliances, changes and review mechanisms, and other incidental matters. The board must approve any subsequent changes made to the model before incorporating them into the policy.
The REs are at liberty to develop the model in-house, engage a third party under contractual obligations, or collaborate. If a third party is engaged, it needs to be reflected in the policy accordingly. The engagement and framework can be supervised by the RBI, either directly or indirectly.
What is a Credit Risk Model?
It is a designated model to assess and ascertain potential risks associated with REs' credit-related activities. The model processes data using financial, statistical, and mathematical principles and projects results for well-informed credit decisions. These decisions may include, but are not limited to, money lending, loan pricing analysis, and estimating capital loss.
The RBI has specified the following broad principles for the model framework:
Clarity: The model should explicitly outline the objectives, problems, and solutions to combat such issues.
Effectiveness and Consistency: The solutions laid out in the model must consistently align with the identified problems. The mechanism must be effective and robust without deviating from the specific issues it aims to resolve.
Detailed Documentation: Detailed documentation should be provided, explaining the sensitivity of the output relative to the input and assumptions to aid user comprehension.
Interface: RE's interface must be consistent with that of other banking/financial entities and with that of its other risk management systems.
Outcomes: The outcomes produced by the model must be consistent, unbiased, explainable and verifiable.
Subjectivity: Subjective factors that may override any model outcome should be considered during policy formulation and incorporated into the model accordingly in an auditable format.
Validation Exercise
The validation process must be extensive and conducted with due diligence. All data, statutory requirements, and compliance aspects should be thoroughly examined, along with the efficacy of the model and outcomes through back-testing. The limitations must be ascertained and corrected accordingly. The solutions to the limitations and their corrections must be conveyed to the board or any committee designated by the board.
If the RE sources its model from a third party, it must be thoroughly vetted to ensure it operates correctly and meets the intended purpose. The model will undergo annual supervisory evaluations to validate its functionality and robustness.
Review by RBI
The models developed by the REs will be supervised by the RBI or external experts, as appropriate.
Our Analysis
In conclusion, the new draft guideline underscores the importance of enhancing the credit risk management framework of the REs. With the new mandates revising the principles of Credit Risk Models emphasising clarity, effectiveness and documentation, the RBI aims to strengthen the immunity of the financial system and build resilience against credit-related risks. The REs have the flexibility to develop their models combined with the validation and backtesting to ensure an efficacious mechanism. The principles propounded by the RBI will ensure the preservation of the economic capital of the REs.
End Note
[i] RBI/2024-25/ DOR.STR.REC./21.04.048/2024-25 dated 05.08.2024.
Authored by Vanshika, Advocate at Metalegal Advocates. The views expressed are personal and do not constitute legal opinions.
More Insights
24-07-2026
10
min read
‘Shylockian’ Lending is ‘Squeezing of Blood’: NCLT Moves Beyond Debt & Default to Reject a Section 7 Application
In a rare departure from the conventional debt-and-default enquiry under section 7 of the IBC, the NCLT, Kochi Bench rejected the financial creditors' petition after characterizing the underlying arrangement as a ‘Shylockian system’ of lending. This court ruling discusses Shylockian lending and examines the strength of the Tribunal's focus on the economic substance of the transaction against established legal principles governing admission under section 7 of the IBC.
10-07-2026
7
min read
Faceless Reassessment after S. 147A: What the Supreme Court Did – and Did Not – Decide
The Supreme Court's decision in Tej Pratap Singh does not settle the JAO–FAO controversy. Following Parliament's retrospective insertion of s. 147A, it remands the issue to the High Courts for fresh consideration. Faceless reassessment was never merely about moving tax files from paper to portal; it fundamentally changed the statutory authority responsible for communicating with the taxpayer, examining the record, drafting the order and completing the assessment. The real question now is how far a retrospective legislative clarification can go.
22-06-2026
8
min read
Claim Admission is not Debt Acknowledgement: Supreme Court on RP’s Role & Limitation under the IBC
Can admission of a claim by a Resolution Professional extend limitation under section 18 of the Limitation Act? In Shankar Khandelwal v. Omkara Asset Reconstruction Pvt. Ltd., the Supreme Court answered this question in the negative, holding that claim admission during CIRP is merely a statutory claim-verification process and not an acknowledgement of debt. The ruling clarifies the RP’s non-adjudicatory role and reinforces important principles governing limitation under the IBC.