RBI ECL norms: Key changes for banks from April 2027

What the RBI changed on April 27, 2026

The Reserve Bank of India (RBI) on April 27, 2026 issued amendments to the framework for asset classification, income recognition and provisioning for commercial banks. The centrepiece is a shift to an Expected Credit Loss (ECL) based approach for provisioning, replacing the earlier incurred-loss method. The Directions are slated to come into force from April 1, 2027, replacing existing norms.

The change matters because provisioning will now be built on a forward-looking view of credit risk, rather than waiting for losses to be incurred. At the same time, the RBI has retained the existing non-performing asset (NPA) classification norms, including the familiar 90-days past due trigger for term loans.

From incurred loss to Expected Credit Loss

Under the new framework, banks must estimate “expected credit loss” as a weighted average of credit losses under different scenarios, with scenario probabilities acting as weights. This design explicitly requires banks to reflect a range of possible outcomes instead of relying on a best-case or worst-case outcome.

Banks must assess at each reporting date whether the credit risk on a financial instrument has increased significantly since initial recognition. If credit risk has not increased significantly, the bank recognises a 12-month ECL. If credit risk increases significantly, the bank must recognise lifetime ECL.

The framework also requires banks to incorporate forward-looking information, including borrower-specific factors and macroeconomic variables, when estimating expected losses.

The three-stage approach to credit risk

The Directions introduce a three-stage classification system that links provisioning directly to deterioration in credit risk:

• Stage 1: Assets with no significant increase in credit risk, requiring 12-month ECL
• Stage 2: Assets with significant increase in credit risk, requiring lifetime ECL
• Stage 3: Credit-impaired assets, requiring lifetime ECL

A rebuttable presumption applies that credit risk has increased significantly when contractual payments are more than 30 days past due. Banks may rebut this presumption using reasonable and supportable information.

A financial instrument is considered credit-impaired when it meets conditions similar to NPAs, including borrower distress, concessions, or a high probability of insolvency.

NPA rules stay, and “default” aligns with NPA tagging

Even as the ECL framework takes over provisioning, the RBI has retained the existing NPA classification norms. A financial asset will continue to be classified as NPA if interest or principal remains overdue for more than 90 days in the case of term loans, or if accounts are “out of order” in overdraft or cash credit facilities, among other conditions.

The Directions also align the definition of “default” with NPA classification, and require borrower-level tagging for asset classification. This borrower-level approach means if one exposure becomes NPA, all exposures to that borrower will be treated as NPA.

PD, LGD and EAD become mandatory building blocks

For ECL computation, banks are required to use a framework based on three key parameters:

• Probability of Default (PD)
• Loss Given Default (LGD)
• Exposure at Default (EAD)

The estimates must be probability-weighted and reflect a range of possible outcomes. The Directions require banks to compare the risk of default at the reporting date with the risk at initial recognition, using reasonable and supportable information.

This shifts the operational focus to measurement discipline, data depth, and consistent segmentation, because changes in credit risk since origination drive stage allocation and provisioning.

Forward-looking scenarios and macro variables

The RBI framework explicitly requires the incorporation of forward-looking information, including macroeconomic variables, into ECL estimates. Multiple scenarios must be used, and the expected loss is the probability-weighted outcome across those scenarios.

This approach is intended to make provisions more responsive to emerging stress, not just realised defaults. It also raises the bar on documentation and explainability, because banks need to show why chosen scenarios, probabilities, and overlays are “reasonable and supportable”.

Effective Interest Rate (EIR) method for income recognition

The Directions mandate adoption of the Effective Interest Rate (EIR) method. Under EIR, the effective interest rate is used to discount estimated future cash flows over the expected life of the financial instrument, aiming for a more accurate measure of income.

All loans outstanding as on March 31, 2027 are required to transition to the EIR regime by March 31, 2030. Separately, the material provided also refers to a transition period until March 31, 2031 for full compliance with the overall ECL shift, indicating that banks will have a phased runway for implementation.

Prudential floors as a regulatory backstop

The RBI has prescribed prudential floors across loan categories and stages, acting as a regulatory backstop to ECL estimates. These floors ensure provisioning does not fall below specified minimum levels even if a model produces a lower number.

In practice, this creates a two-layer safeguard: model-based ECL anchored in forward-looking risk assessment, and minimum provisioning levels that prevent under-provisioning during benign periods.

Governance, model validation and data controls

The framework requires banks to establish robust governance structures for ECL implementation. A Board or Board-approved committee, including the Chief Financial Officer (CFO) and Chief Risk Officer (CRO), is required to oversee implementation, ensure consistency in methodologies, and maintain data integrity.

Banks must also implement model risk management frameworks, including validation, monitoring, and independent oversight across the lifecycle of ECL computation. This places explicit regulatory expectations around data management, model inventories, and controls.

Transition mechanics: fair value adjustment through reserves

On transition to the ECL framework from April 1, 2027, banks are required to fair value their entire loan portfolio. Any difference between fair value and carrying amount will be adjusted against retained earnings and not through the profit and loss account.

This accounting treatment concentrates the initial transition impact in reserves rather than in the reported profit for the period, while still requiring banks to do the work of fair valuation across the portfolio.

Summary table: the key elements at a glance

Market impact and why this matters

The ECL approach changes the timing of loss recognition by requiring banks to build provisions based on expected losses and changes in credit risk, rather than waiting for a default event to occur. For investors and analysts, this can alter the pattern of provisioning across the credit cycle, especially where Stage 2 migrations increase lifetime ECL requirements.

The retention of NPA norms alongside ECL is a key design choice. It keeps supervisory triggers and classification conventions intact, while provisioning becomes more sensitive to early-warning signals such as significant increases in credit risk and probability-weighted scenario outcomes.

Implementation requirements on governance, model validation, and data management also indicate tighter expectations on how banks justify assumptions, validate models, and control end-to-end ECL computation.

What to watch as April 2027 approaches

Between now and the April 1, 2027 start date, key operational milestones for banks will include building staging logic, developing PD-LGD-EAD frameworks, embedding macroeconomic scenarios, and setting up governance structures that meet the RBI’s oversight expectations.

Another focus area will be the migration to EIR-based income recognition for loans outstanding as of March 31, 2027, within the timelines specified in the Directions. Market participants are likely to track disclosures around transition adjustments made through retained earnings and the evolution of provisioning under the prudential floors.