RBI Rolls Out Regulatory Principles For Managing Model Risks In Credit

The Reserve Bank of India (RBI) proposed some new rules for managing the risks associated with algo-based lending. Regulated Entities (REs) utilise models throughout the credit management life cycle for borrower selection, credit scoring, pricing, risk management, and credit loss provisions. The draft circular from RBI, ”Regulatory Principles for Management of Model Risks in Credit”, provides guidelines for lenders to address uncertainties in model outputs. According to the RBI, credit risk models use quantitative methods—such as statistical, economic, financial, or mathematical assumptions—to analyze data for credit decisions. The RBI’s draft circular directs all lenders to establish a detailed, board-approved policy for managing model risk across all used models. This policy mandates that any changes to inputs in individual credit models must be approved by the board’s risk management committee or another designated committee.

Algo-based lending:

Algo-based lending employs advanced mathematical models and algorithms to evaluate creditworthiness, assess lending risk, and decide on loan approvals, utilising extensive data to predict borrowers’ repayment likelihood.

RBI’s Draft:

This potentially exposes the REs (banks, NBFCs and housing finance companies) to model risk, which has implications on prudential aspects of credit risk management, compliance and reputational risk, the RBI further stated. The RBI stated that regulated entities must create a board-approved policy for all risk management model frameworks they use, including justifications for adopting third-party models.

All individual credit model deployment and subsequent changes would be based on the approval of the risk management committee of the entity’s board, the RBI stated. New credit models to be adopted by REs shall follow these guidelines with effect from the above date and the existing models shall be validated in terms of these guidelines within six months from the date of issuance of this circular. The regulations follow the RBI’s recent warning to NBFCs about excessive reliance on algorithm-based credit models.

Guidelines propose from the RBI to all the lenders:

Lenders may use models developed inhouse or sourced from third-party suppliers, including collaborative arrangements or a combination of both. All models must be thoroughly documented, detailing the sensitivity of their outputs, and should be scalable and flexible to adapt to changing business conditions. They need an interface with core banking systems, liquidity management, asset-liability management, or other risk management functions. Additionally, lenders must have an independent model evaluate and validation process to evaluate the robustness of both in-house and external models. Models deployed by lenders should undergo supervisory review, with the RBI potentially engaging external experts for validation. The draft guidelines would come into effect within three months from the date of its issue and all new credit models that will be adopted by lenders must follow them.

Conclusion: RBI is tightening the algorithm-based lending process to eliminate loopholes, ensuring that it is trackable and easier for auditing.

Disclaimer: This blog has been written exclusively for educational purposes. The securities mentioned are only examples and not recommendations. It is based on several secondary sources on the internet and is subject to changes. Please consult an expert before making related decisions.