Expected Credit Loss Under IFRS 9: The Three-Stage Model Explained

Expected credit loss accounting changed impairment from a backward-looking to a forward-looking discipline. Under IFRS 9, you recognise losses you expect to incur, not just losses that have already happened. This guide explains the three-stage model, the simplified approach for receivables, and how to build defensible ECL estimates.

Why did IFRS introduce expected credit losses?

The expected credit loss model was a direct response to the 2008 financial crisis, when the previous incurred-loss approach was blamed for recognising loan losses too late — only after a default event was evident. By then, the damage was done and balance sheets had overstated asset values for years. The ECL model forces earlier recognition based on anticipated losses, smoothing the recognition of credit risk over time.

The shift is philosophical as well as technical. It asks preparers to look forward, incorporating macroeconomic forecasts and reasonable expectations, rather than waiting for losses to materialise. This is more demanding but produces a more realistic picture of credit risk on the balance sheet.

How does the three-stage model work?

The general model assigns every in-scope asset to one of three stages. Stage 1 covers performing assets, where you recognise twelve-month expected losses — the losses expected from default events possible within the next year. When credit risk has increased significantly since initial recognition, the asset moves to Stage 2 and you recognise lifetime expected losses. Once the asset is credit-impaired, it enters Stage 3, where lifetime losses are recognised and interest is calculated on the net carrying amount.

The trigger between Stage 1 and Stage 2 — a significant increase in credit risk — is one of the most judgmental aspects. Entities define indicators such as days past due, rating downgrades, or covenant breaches. Where an asset sits determines how much loss you carry, so the staging logic must be documented and applied consistently.

What inputs do you need to calculate ECL?

A full ECL calculation combines three core inputs: the probability of default (how likely the borrower is to default), the loss given default (how much you lose if they do, after recoveries), and the exposure at default (how much is outstanding when default occurs). These are multiplied and discounted to present value, then weighted across multiple economic scenarios.

Gathering these inputs is the hard part for corporates, who rarely have the historical default databases that banks maintain. Practical solutions include external credit ratings, published default statistics, and credit insurance data. Whatever the source, the methodology must be documented and the forward-looking adjustments must be explainable.

How does the simplified approach for trade receivables work?

Recognising that the full three-stage model is overkill for short-term trade receivables, IFRS 9 permits a simplified approach. You build a provision matrix grouping receivables by age bracket and apply a historical loss rate to each bracket, then adjust those rates for current and forecast economic conditions. The result is a lifetime expected loss without the need to track significant increases in credit risk.

For most non-financial businesses, this is the workhorse method. The key is to base the loss rates on genuine historical experience and to make the forward-looking adjustment defensible — for instance, increasing rates when economic conditions are deteriorating in a key customer market.

How do economic scenarios feed into ECL?

IFRS 9 requires ECL to reflect a range of possible outcomes, not a single best estimate. In practice, entities model several macroeconomic scenarios — typically a base case plus optimistic and pessimistic variants — and weight them by probability. GDP growth, unemployment, interest rates, and sector-specific indicators feed into the default and loss assumptions.

This scenario weighting introduces volatility and judgment into the impairment number, which can move significantly as economic forecasts change. During periods of uncertainty, ECL provisions can swing sharply, and management must be ready to explain those movements to stakeholders as forecast-driven rather than reflecting actual defaults.

How should corporates govern their ECL process?

Because ECL is judgmental and material, it needs proper governance. Leading practice is to document the methodology, define the staging triggers, approve the economic scenarios and weights at a senior level, and review management overlays through a formal committee. The process should be repeatable each period and produce an audit trail that an auditor can follow without bespoke explanation.

For groups, governance also means consistency across entities — the same staging logic, scenario set, and methodology applied group-wide, so that the consolidated ECL is coherent. This connects to the broader discipline of consistent group accounting policy explored across our IFRS hub.

How does ECL affect intercompany loans within a group?

A frequently missed application of the expected credit loss model is intercompany lending. In the separate financial statements of group entities, intercompany loans are financial assets subject to ECL just like third-party loans. Even where repayment seems assured because the borrower is a fellow subsidiary, IFRS 9 requires an assessment of credit risk and recognition of expected losses, particularly for loans repayable on demand or to entities with limited liquid resources.

This catches many groups off guard, especially for long-dated or undocumented intercompany funding. The assessment must consider the borrower’s ability to repay if the loan were called, which can require recognising a loss even on a loan the parent never intends to demand. While these losses often eliminate on consolidation, they affect the standalone statutory accounts that drive local filing and distributable reserves, linking back to the dual-reporting discipline explored in our IFRS hub.

How do you build ECL when you lack historical default data?

Most non-financial companies do not have the rich internal default databases that banks rely on, yet they still must produce a defensible expected credit loss estimate. Practical sources fill the gap: published default and recovery statistics by rating or sector, external credit ratings and scores, credit insurer data, and the entity’s own collection and write-off history for trade receivables. The methodology need not be sophisticated, but it must be evidenced and consistently applied.

The key is proportionality. A modest trade-receivables book can be adequately covered by a well-constructed provision matrix grounded in real collection experience, with a transparent forward-looking overlay. Over-engineering the model is as much a risk as under-supporting it; the goal is an estimate that faithfully represents expected losses and that an auditor can follow and accept.

How did economic volatility test the ECL model?

The expected credit loss model has been stress-tested by real-world volatility since its introduction. Periods of economic shock force entities to rapidly update their forward-looking scenarios, often increasing provisions sharply before any actual defaults occur — exactly the early recognition the model was designed to produce. The flip side is that ECL provisions can be volatile and judgmental, swinging on changes in forecasts rather than realised losses, which management must explain carefully to stakeholders.

This volatility is a feature, not a bug, but it places a premium on clear communication. When provisions rise because the economic outlook has darkened, that is the model working as intended; when they fall as conditions improve, the reversal is equally valid. Finance teams that frame ECL movements as forward-looking risk signals, distinct from actual credit performance, help investors and lenders interpret the numbers correctly rather than reading every provision swing as deteriorating loan quality.

What is the role of write-offs and recoveries in ECL?

Expected credit loss accounting governs the provision, but actual write-offs and recoveries are separate events that must also be handled correctly. A write-off occurs when there is no reasonable expectation of recovering all or part of a financial asset; it reduces the gross carrying amount and is applied against the previously recognised loss allowance. Subsequent recoveries of amounts written off are recognised when received. The provision, the write-off, and the recovery are three distinct steps that together tell the full credit-loss story.

Keeping these distinct matters for both accuracy and analysis. The loss allowance reflects expected future losses; the write-off reflects a realised, unrecoverable loss; and recoveries reflect cash collected against earlier losses. Conflating them obscures the true credit performance of the portfolio. A disciplined process tracks each separately and reconciles the loss allowance movement each period, which is also what the IFRS 7 disclosures require.

How does ECL reporting build credibility with lenders and investors?

A well-governed expected credit loss process does more than satisfy the auditor; it builds credibility with the lenders and investors who rely on the balance sheet. When a company can demonstrate a documented methodology, evidenced loss rates, transparent economic scenarios, and governed management overlays, stakeholders gain confidence that the impairment number reflects genuine risk rather than convenient estimation. That confidence translates into easier financing conversations and a lower risk premium.

The opposite is equally true: opaque or volatile provisions that management cannot explain erode trust quickly. Treating ECL as a communication discipline as much as a calculation — explaining why provisions moved, distinguishing forecast effects from actual defaults, and disclosing the key assumptions — turns a technical requirement into a source of credibility, consistent with the transparency theme across our IFRS hub.

Frequently Asked Questions