PD, LGD and EAD: Turning Credit Risk into Expected Credit Loss

Probability of default, loss given default and exposure at default are widely used building blocks for ECL, but Ind AS 109 does not make the three-factor formula a substitute for the standard's measurement objective. The components must collectively produce probability-weighted, discounted expected cash shortfalls over the relevant horizon. The strongest accounting files make the reasoning visible, so that a reviewer can understand not only the conclusion but also why plausible alternatives were rejected. Component models can look statistically strong while still producing the wrong accounting result if definitions, timing and interdependencies are inconsistent. A robust approach connects commercial substance, the Ind AS 109 decision criteria, measurement evidence and presentation consequences in one coherent file.

Define default consistently

The core requirement. PD depends on a clear default definition that is aligned across risk systems, staging, recoveries and validation while reflecting the accounting purpose. In a controlled close process, entities should reconcile regulatory or internal default definitions to the definition used in ECL and explain justified differences. A common weakness is calibrating PD on one default event while LGD and cure data are built on another. The accounting result should reconcile to the underlying contract, valuation or subledger rather than rely on a standalone spreadsheet conclusion.

Build point-in-time risk into PD

The principle. ECL requires current and forward-looking information, so purely long-run or through-the-cycle default rates usually need adjustment before they represent reporting-date expectations. For a review-ready file, methodologies may use transition models, macroeconomic linkages or other techniques, but the transformation should be empirically supported and governed. The risk to avoid is applying opaque management scalars to TTC PDs without explaining the economic relationship. The working paper should identify the relevant facts, source data, judgement and conclusion so that an independent reviewer can reproduce the decision.

Measure LGD as an economic shortfall

The technical anchor. LGD should capture expected recoveries, costs, cure behaviour, collateral realisation and the timing of cash flows rather than only historical write-off percentages. In application, secured portfolios need realistic haircuts, workout periods and recovery costs, with discounting consistent with the ECL framework. A frequent failure mode is using current collateral values at face amount and ignoring time to recovery. Where the conclusion is sensitive to a contractual clause or estimate, the file should show the alternative outcome and why the selected treatment is more appropriate.

Estimate EAD over the relevant horizon

The accounting logic. EAD should reflect expected outstanding exposure when default could occur, including amortisation, repayments, prepayments and expected drawings for commitments where relevant. Operationally, cash-flow schedules and credit-conversion assumptions should be consistent with the contractual product and observed borrower behaviour. The main judgement risk is holding today's balance constant through lifetime ECL when the exposure is expected to change materially. A concise review note should state the trigger, the rule applied, the evidence considered and the financial-statement consequence.

Control interactions and aggregation

The decision point. PD, LGD and EAD are not always independent, especially in stressed conditions when defaults rise, collateral values fall and utilisation increases. For implementation, scenario design and model validation should test whether correlations and segmentation capture these relationships adequately. Where errors often arise is combining individually reasonable components in a way that understates joint stress. Evidence should be retained at the same level of detail as the accounting conclusion, with assumptions version-controlled and exceptions explicitly approved.

Practical illustration

Consider a revolving facility where stressed borrowers tend to draw available limits before default. Using the current drawn balance as EAD can understate expected exposure even if PD and LGD are well calibrated. The model needs an evidence-based drawdown assumption that interacts appropriately with the default horizon and the institution's ability to cancel the facility. The illustration is deliberately simplified: its purpose is to show how the accounting conclusion follows the underlying facts rather than to prescribe a single mechanical answer for every entity. Before posting an entry, the preparer should reconcile contractual terms, management's commercial intent, relevant estimates and system data to the specific accounting requirement. Where the outcome is sensitive, the file should show the key judgement and explain why the selected assumption is reasonable at the reporting date.

Documentation and control points

Professional application depends as much on process quality as technical knowledge. For this topic, a minimum control set should cover default-definition alignment; model calibration and validation; collateral data quality; drawdown modelling; and component-to-ECL reconciliation. Ownership should be clear between the business, finance and any legal, tax, valuation, credit-risk or other specialists whose evidence is required. Source data should be dated and version-controlled; manual adjustments should show preparer, reviewer, rationale and approval. The final accounting memorandum should connect the conclusion to the general ledger or relevant subledger, presentation and disclosures. If facts or estimates change, the entity should reassess the conclusion when required and preserve an audit trail explaining the change.

Closing perspective

PD, LGD and EAD are useful only when their definitions and timing converge on the same expected cash-shortfall objective. The most useful way to apply Ind AS 109 is to treat the requirement as a decision framework rather than a compliance slogan. When the facts, accounting criteria, measurement evidence, controls and disclosure implications are considered together, the result is more consistent across reporting periods and easier to explain to management, auditors and users of the financial statements.