1. Introduction: Why ECL?

For decades, Indian banks operated under the Incurred Credit Loss (ICL) model, where provisions were made only after a loan had actually deteriorated into a Non-Performing Asset (NPA). This backwards-looking approach created a fundamental problem: banks recognised losses too late, leading to sudden and large provisioning hits that shocked earnings and eroded capital precisely when the economy was already under stress. The Reserve Bank of India, through its ACPIR Directions, 2026 (RBI / DOR/2026-27 / 398, dated April 27, 2026), has replaced this reactive model with the Expected Credit Loss (ECL) framework, a forward-looking and probability-weighted approach that requires banks to estimate potential losses from the very first day a loan is originated and to continuously update that estimate as credit conditions evolve.

The framework, effective from April 1, 2027, introduces three landmark changes: i) a three-stage asset classification based on credit risk evolution; ii) forward-looking provisioning using macroeconomic scenarios; and iii) the Effective Interest Rate (EIR) method for income recognition. Together, these changes transform not just how banks provision, but how they manage credit risk, disclose financial information and plan capital.

The ECL framework has a direct and material impact on three critical dimensions of a bank's financial health: provisioning requirements increase across the board; capital adequacy comes under pressure from higher provision charges; and profitability becomes more sensitive to early credit risk signals. Understanding these linkages is the central objective of this article.

2. What is Expected Credit Loss?

ECL is the ‘weighted average of credit losses under different scenarios, with the respective probability of the various scenarios as the weights.’ In operational terms, ECL is computed using three building blocks, which are explained in Table 1.

ECL = PD x LGD x EAD

A critical regulatory constraint is the minimum PD floor of 0.03 percent: no instrument can carry a PD below this threshold, regardless of model output. For LGD, where a bank cannot estimate internally, regulatory backstop values apply: 65 percent for the secured portion and 70 percent for the unsecured portion.

3. The three-stage classification framework

Every financial instrument is classified into one of three stages based on changes in credit risk since initial recognition. The stage determines both the time horizon for ECL measurement and the minimum provision (prudential floor) the bank must hold. This is the architectural backbone of the ECL framework. Table 2 shows the three-stage classification framework. The prudential floors applicable to key loan product categories across all three stages are shown in Table 3.

Two observations are critical for capital management. First, the Stage 2 jump from 5 percent of EAD is not a modelling output: it is a regulatory minimum. A bank's own model may compute a lower figure, but it cannot provide below this floor. Second, the Stage 3 escalation is time-driven, not model-driven: regardless of collateral valuation or recovery prospects, the floor rises automatically, forcing banks to build provisions progressively rather than in a single shock.

Stage 3 classification is applied at the borrower level: if any one exposure of a borrower is Stage 3, all exposures to that borrower, including non-funded exposures, are classified as Stage 3. Stage 2 classification is applied at the facility level.

4. Significant Increase in Credit Risk (SICR)

SICR is a significant or material change in the estimated default risk over the remaining expected life of the financial instrument. It is the trigger for migration from Stage 1 to Stage 2 and, with it, the shift from 12-month ECL to lifetime ECL. Understanding SICR is therefore central to understanding when and how provision costs surge.

4.1 Quantitative triggers: Hard rules

RBI prescribes a rebuttable presumption of SICR through objective, system-driven thresholds:

  • For term loans and bills: payments more than 30 Days Past Due (DPD).
  • For OD / CC revolving facilities: outstanding balance continuously in excess of sanctioned limit or drawing power, whichever is lower, for more than 60 days.
  • Internal credit rating downgraded by a defined number of notches as per the bank's Board-approved policy.
  • Significant increase in loan pricing attributable to higher credit risk.

These are rebuttable presumptions, not absolute rules. A bank may rebut the SICR presumption for a specific account if it has strong, documented evidence that credit risk has not actually increased, for instance, a seasonal business with a historically perfect track record experiencing a one-time delay.  However, such rebuttal must be board-approved, consistently applied, and cannot be used routinely or mechanically.

4.2 Qualitative indicators: Judgement-based signals

Beyond the quantitative thresholds, bankers must monitor early warning signals and trigger SICR proactively:

  • Significant decline in borrower revenues, margins, or asset quality.
  • Deterioration in borrower's leverage, liquidity, or working capital position.
 
To read more, please subscribe.