Predicting the Unpredictable: Credit Loss Provisioning Under Ind-AS 109

The Forward-looking Expected Credit Loss under Ind-AS 109 represented a regime change in the Loan Loss Provisioning as against the reactive approach as prescribed under the Accounting Standards regime. Further, the same is also considerably different from the IRACP Norms as prescribed by the Reserve Bank of India. Ind-AS 109 talks about the Expected Credit Loss model instead of the Incurred Credit Loss model. ECL provides the framework not only based upon the past and current information but also expected credit losses based upon past experiences. Ind-AS 109 does not prescribe any methodology for computing ECL, however, entities are expected to provide for the losses in accordance with the asset size, intricacy, and risk profile of the company. Let\'s dive into each aspect of the Loan Loss Provisioning under the ECL Model.

Classification of Financial Assets under Ind-AS 109

Under Ind-AS 109 & Ind-AS 32, asset classification principles hinge on an entity\'s business model for managing financial assets and the contractual cash flow characteristics of the assets. Three main categories for financial asset classification are as below:

i. Amortised Cost

• Criteria: A financial asset is classified at amortized cost if both conditions are met:
  • The asset is held within a business model whose objective is to hold financial assets to collect contractual cash flows.
  • The contractual terms of the financial asset give rise to cash flows that are solely payments of principal and interest (SPPI) on the principal amount outstanding.

• Measurement: Measured at amortized cost using the effective interest rate (EIR) method, with impairment losses recognized as needed.

ii. Fair Value Through Other Comprehensive Income (FVOCI)

• Criteria: A financial asset is classified at FVOCI if both conditions are met:
  • The asset is held within a business model whose objective is achieved by both collecting contractual cash flows and selling financial assets.
  • The contractual terms of the financial asset give rise to cash flows that are SPPI.

• Measurement: Changes in fair value are recognized in Other Comprehensive Income (OCI), except for impairment gains or losses and interest revenue, which are recognized in profit or loss.

iii. Fair Value Through Profit or Loss (FVTPL)

• Criteria: A financial asset is classified at FVTPL if:
  • It does not meet the criteria for classification at amortized cost or FVOCI, or
  • It is designated as an FVTPL upon initial recognition to eliminate or significantly reduce an accounting mismatch.

• Measurement: Measured at fair value, with changes in fair value recognized in Profit or Loss.

Financial Assets and Items Subject to ECL

• Financial Assets measured at Amortised Cost (e.g. Trade Receivables, Loans)
• Financial Assets measured at FVOCI
• Lease Receivables (Ind AS 116)
• Contract Assets (Ind AS 115)
• Loan Commitments not classified as financial liabilities at FVTPL (e.g. Guarantees)
• Financial guarantee contracts (if not designated at FVTPL) require impairment based on the ECL model.

The ECL model does not apply to:

• Equity instruments (measured at FVOCI or FVTPL), as they do not have contractual cash flows.
• Financial assets are measured at FVTPL, since changes in fair value already account for credit risk dynamically.

Approaches for calculation of ECL

Ind-AS 109 does not prescribe how to derive the ECL; the methodology used could vary based upon the type of assets, complexities involved & information available at the time of the provisioning. Here\'s the overview of the approaches that can be used:

General Approach

The General Approach is applied to most financial assets subject to ECL unless they qualify for the simplified approach. It involves a 3-stage impairment model:

• Stages of Credit Risk and ECL Recognition:
  1. Stage 1 (Performing Assets):
  2. Stage 2 (Underperforming Assets)
  3. Stage 3 (Credit-Impaired Assets)
• Key Components of the General Approach:
  • Probability of Default (PD): Likelihood that the borrower will default.
  • Loss Given Default (LGD): Expected loss if a default occurs, as a percentage of exposure.
  • Exposure at Default (EAD): Outstanding amount at the time of default.

• Formula: PD X LGD X EAD X Discounting Factor

Simplified Approach

The Simplified Approach is mandatory for:

• Trade receivables, contract assets, and lease receivables that do not contain a significant financing component.
• Optional for trade receivables and contract assets with a significant financing component.

Features

• No need to track changes in credit risk or classify assets into stages.
• Lifetime ECL is always recognized from initial recognition.
• Typically involves creating a provision matrix that uses historical credit loss experience, adjusted for forward-looking factors (e.g., economic conditions).

Example of a Provision Matrix

Segmentation of the Loan Portfolio

Segmentation involves grouping financial assets with similar risk characteristics to estimate credit losses more accurately and efficiently.

Purpose of Loan Portfolio Segmentation

• Reflects the different levels of credit risk inherent in various types of loans.
• Ensures that ECL estimates are tailored to the specific characteristics of each segment.
• Enhances accuracy by aligning with historical and forward-looking data.

Factors for Segmentation

Segmentation is typically based on the following criteria:

• Type of Borrower
• Type of Loan
• Purpose of Loan
• Geographic or Sectoral Factors
• Risk Attributes like Credit Risk Rating, Loan to Value ratio.

By appropriately segmenting the loan portfolio, entities can enhance the precision of ECL estimates and provide more meaningful financial reporting under Ind AS 109.

Staging

Staging of financial assets under Ind AS 109 refers to categorizing financial assets into three distinct stages based on changes in credit risk since initial recognition. This staging determines how ECL are measured and recognized, aligning impairment provisions with the degree of credit risk.

Stages of Financial Assets

Stage 1: Performing Assets

• Criteria: Financial assets where there has been no significant increase in credit risk (SICR) since initial recognition.
• ECL Measurement
  • 12-month ECL: Losses expected from defaults occurring within 12 months after the reporting date.
  • Reflects the probability of default in the next 12 months, regardless of the expected life of the asset.
• Interest Income Calculation
  • Based on the gross carrying amount (before deducting impairment allowance).

Stage 2: Underperforming Assets

• Criteria: Financial assets with a significant increase in credit risk since initial recognition but not yet credit impaired.
• ECL Measurement
  • Lifetime ECL: Losses expected over the entire remaining life of the asset.
  • Assesses the likelihood of default over the asset\'s full term.
• Interest Income Calculation
  • Based on the gross carrying amount (before impairment).

Stage 3: Credit-Impaired Assets

• Criteria: Financial assets that are credit-impaired, typically involving one or more of the following:
  • Significant financial difficulty of the borrower.
  • Breach of contract, such as a default or past-due event.
  • High probability of bankruptcy or financial reorganization.
  • Concessions granted to the borrower due to financial difficulty.
• ECL Measurement
  • Lifetime ECL: Calculated based on expected losses for the asset\'s remaining life.
  • Incorporates detailed, borrower-specific information.
• Interest Income Calculation
  • Based on the net carrying amount (gross carrying amount minus impairment allowance).

Staging & ECL Recognition

Significant Increase in credit risk

A Significant Increase in Credit Risk (SICR) under Ind AS 109 occurs when the credit risk of a financial instrument has risen substantially since its initial recognition. SICR is a critical trigger for reclassifying financial assets from Stage 1 (Performing) to Stage 2 (Underperforming) under the Expected Credit Loss (ECL) model.

Indicators of SICR

SICR assessment combines quantitative, qualitative, and backstop measures. These include:

• Quantitative Indicators
  • Changes in Probability of Default (PD)
  • Credit Rating Downgrades
• Qualitative Indicators
  • Adverse Changes in Financial Conditions
  • Operational or Sectoral Risks
  • Forbearance or Restructuring Measures
  • Adverse Business Developments
• Backstop Indicators
  • Days Past Due more than 30 days
  • Regulatory Triggers

Assessing SICR

SICR assessment involves:

• Comparison to Initial Recognition
• Forward-Looking Information
• Rebuttable Presumptions

Examples of SICR

Probability of Default (PD)

Probability of Default (PD) is a key component of the Expected Credit Loss (ECL) model. It measures the likelihood that a borrower will default on their financial obligation over a specified time horizon. PD is essential for estimating credit losses and reflects both historical experience and forward-looking factors. PD is the probability that a counterparty will fail to meet its debt obligations, leading to a default event.

Stages and Use of PD

PD Estimation Techniques

PD can be estimated using a combination of the following methods:

• Historical Data Analysis:
  • Analyse historical default rates for borrowers or similar risk groups.
  • Use internal or external credit risk rating systems to estimate default likelihood.

• Logistic Regression
  The Logistic Regression method is a widely used statistical technique for modelling the Probability of Default (PD) in credit risk analysis. It is particularly effective because it predicts probabilities (values between 0 and 1) and handles binary outcomes, such as whether a borrower defaults (1) or does not default (0).
  
  The model estimates the probability of default, expressed as:
  
  $$ P(Default) = \\frac{1}{1 + e^{-(\\beta_{0} + \\beta_{1}X_{1} + ... + \\beta_{n}X_{n})}} $$
  
  Where:
  $\\beta_{0}, \\beta_{1}, ..., \\beta_{n}$ are the coefficients (to be estimated).
  $X_{1}, X_{2}, ..., X_{n}$ are the predictor variables.
• Proportional Hazard Models
  Proportional Hazard Models (PHM), particularly the Cox Proportional Hazard Model (Cox Model), are another sophisticated approach used for estimating Probability of Default (PD) in credit risk modelling. These models are based on survival analysis, which is typically used to analyse time-to-event data, such as the time until a borrower defaults. The Proportional Hazard Model is specifically useful when the focus is on estimating the hazard rate (the risk of default) as a function of covariates (borrower-specific characteristics, economic factors, etc.). The model assumes that the hazard rate (default risk) at any time is a function of baseline hazard and covariates.
  
  The hazard function for an individual i is given by:
  
  $$ h(t|X_{i}) = h_{0}(t)exp(\\beta_{1}X_{i1} + \\beta_{2}X_{i2} + \\dots + \\beta_{n}X_{in}) $$
  
  Where:
  $h(t|X_{i})$ is the hazard rate (risk of default at time t for borrower i).
  $h_{0}(t)$ is the baseline hazard function (default risk at time t for an average borrower with $X_{1} = X_{2} = \\dots = X_{n} = 0$).
  $X_{1}, X_{2}, ..., X_{n}$ are the covariates (predictor variables such as borrower-specific characteristics and macroeconomic factors).
  $\\beta_{1}, \\beta_{2}, ..., \\beta_{n}$ are the coefficients estimated by the model.
  
  The hazard rate represents the instantaneous probability of default at any given point in time, conditioned on the borrower surviving up to that time.
• Markov Chains:
  A Markov Chain is a statistical model that describes a sequence of possible events (states) in which the probability of each event depends only on the state attained in the previous event. This makes it particularly useful for modelling stochastic processes like credit risk, where the future state (e.g., whether a borrower defaults) depends on the current state (e.g., the borrower\'s credit rating).
  
  In the context of PD estimation, Markov Chains are used to model the transition probabilities between different credit states (e.g., healthy, overdue, defaulted) over time.
  
  The model divides the credit quality of loans into discrete states, such as:
  
  • Healthy (not yet in default)
  • At Risk (loans that are overdue but not yet defaulted)
  • Defaulted (loans that have defaulted)

Other modelling techniques like Decision Trees, Credit Scoring Models, Bayesian Models, Discriminant Analysis, Merton Model, Machine Learning Models, etc. can also be used in the estimation of PD.

Key Considerations in PD Estimation

Choosing the correct model for PD estimation under the ECL framework is a critical decision for financial institutions. The model must be able to accurately assess the likelihood of default while complying with the regulatory and accounting standards set by IND-AS 109. Here are key steps and factors to consider when selecting a model for PD estimation under ECL:

• Understand the Data Characteristics
  • Data Availability and Quality
  • A model\'s ability to handle multivariate data (numerical and categorical) and temporal features (e.g., time-series data) is important
• Model Complexity and Interpretability
  • Simplicity vs. Complexity
  • Regulatory Requirements - Regulatory standards may demand that models be interpretable, particularly when justifying credit loss provisions
• Segmentation and Risk Characteristics
  • Risk Segmentation
  • Loan Type - Secured / Unsecured

• Forward-Looking Considerations
• Computational and Implementation Constraints
  • Scalability
  • Implementation Costs

• Model Integration with Risk Management Framework

Ultimately, the model chosen should provide an optimal balance of predictive accuracy, regulatory compliance, operational feasibility, and interpretability to ensure that PD estimation is robust, transparent, and adaptable to changing economic conditions.

Loss Given Default (LGD)

Loss Given Default (LGD) refers to the potential loss a lender or investor incurs if a borrower defaults on a loan or credit facility. It is defined as the amount of loss that remains after accounting for any recoveries from the defaulted loan, usually expressed as a percentage of the total exposure at default (EAD).

LGD Estimation Process

• Historical Data
  Use historical loss data to estimate the potential recoveries and default severity. This data might include past loan defaults, collateral liquidations, and recoveries in similar economic conditions.
• Discounted Cash Flow Approach
  If there is an expected recovery over time, the LGD can be calculated using a discounted cash flow (DCF) approach, where future expected recoveries are discounted to the present value.

In summary, Loss Given Default (LGD) is a crucial parameter under Ind-AS 109 for determining the Expected Credit Loss (ECL) for financial assets. The LGD estimation considers factors such as collateral value, recovery costs, loan structuring, and the economic environment. The accuracy of LGD estimation is vital for financial institutions to properly assess the credit risk and ensure the appropriate provisioning for potential losses.

Exposure at default (EAD)

Exposure at Default (EAD) represents the total value of a financial asset or portfolio at the time of a borrower\'s default. EAD helps estimate the potential financial loss an entity might incur due to credit defaults, which is essential for determining appropriate provisions for credit risk.

EAD = Loan Balance + Accrued Interest + Any Undrawn Commitments

Where:

• Loan Balance is the current outstanding loan amount.
• Accrued Interest is the interest that has been accumulated but not yet paid.
• Undrawn Commitments represent any unused portion of a revolving credit facility (e.g., unused credit on a credit card or line of credit) that could be drawn by the borrower before default.

Other Aspects to be kept under consideration while calculating ECL

In addition to the key components such as PD, LGD, and EAD, there are several other aspects that need to be considered while calculating ECL under Ind-AS 109. These include, the use of forward-looking information, macroeconomic factors, and credit risk mitigation factors, among others. Below are the important additional considerations to be kept in the mind while calculating ECL under Ind-AS 109:

• Forward-Looking Information
  Under Ind-AS 109, the calculation of ECL requires the use of forward-looking information. This means that the financial institution should not only rely on historical data but should also take into account expected future conditions that may affect credit risk. These conditions can be macroeconomic factors or any other relevant information that might impact the borrower\'s ability to repay.
  • Macroeconomic scenarios: Variables like GDP growth, inflation rates, unemployment levels, and interest rates can have a significant impact on the creditworthiness of borrowers and on the ECL estimation.
  • Scenario Analysis: Financial institutions are encouraged to consider a range of scenarios, including base, optimistic, and pessimistic scenarios, when determining the PD, LGD, and EAD. This helps to adjust the ECL for both positive and negative future outcomes.
  • Historical information: Historical data, while still valuable, must be adjusted to reflect the anticipated future conditions in line with the financial institution\'s internal credit models.
• Use of Credit Risk Mitigation (CRM) Techniques
  When calculating ECL, institutions should consider the impact of any credit risk mitigation (CRM) techniques, such as collateral or guarantees. These techniques reduce the potential loss given default (LGD), thus affecting the overall ECL.
  • Collateral: The value of collateral should be assessed regularly, and its potential recovery value should be taken into account while estimating LGD. Collateral types, such as real estate, cash, or receivables, will influence the recovery rate.
  • Guarantees: If there are third-party guarantees (e.g., corporate guarantees, government guarantees, or personal guarantees), the potential recovery from these guarantees should be considered in the calculation of LGD and, subsequently, the ECL.
  • Netting off: In some cases, institutions may offset liabilities or recoveries from collateral against the gross exposure, depending on the applicable legal framework and contractual terms.
  • Cross-collateralization: If multiple exposures are backed by the same collateral, the impact of such arrangements should be considered when calculating the ECL.

• Accounting for Prepayments and Repayments
  When calculating ECL for loans that are subject to prepayment (e.g., mortgages), financial institutions must account for the possibility that the loan may be repaid before default. As the prepayment reduces the EAD while. Also, for the amortizing loans or revolving credit facilities expected repayment schedules shall be considered as Balance and exposure might change over the time.
• Loan Modifications and Restructuring
  In cases of loan modifications or restructuring (e.g., forbearance, repayment extensions, or changes in loan terms), the institution should reassess the credit risk of the asset and its corresponding ECL.
  • Modification impact: Loan modifications might reduce or increase credit risk. If the modification results in a significant increase in credit risk, the asset may move to Stage 2 or 3.
  • Default assessment after modification: After a restructuring, it is important to reassess the asset\'s credit risk and adjust the PD and LGD based on the new terms.

In conclusion, the calculation of Expected Credit Loss (ECL) under Ind-AS 109 is a comprehensive process that requires a detailed assessment of credit risk across different stages of a financial asset\'s life cycle. It involves not only key components like PD, LGD, and EAD but also the integration of forward-looking information, macroeconomic factors, and credit risk mitigation strategies. The application of a forward-looking, probabilistic approach ensures that provisions for credit losses remain reflective of both current and anticipated market conditions. By considering factors such as staging, loan modifications, and the impact of prepayments, financial institutions can arrive at an accurate and realistic ECL estimation, which helps maintain sufficient reserves against potential credit losses, thus enhancing financial stability and regulatory compliance.