Understanding the Incurred Loss Model: Key Insights

1. Why this topic matters for banks, lenders and reporting entities

The choice of impairment approach affects your balance sheet, provisioning volatility, capital planning and stakeholder communication. While IFRS 9 requires forward‑looking Expected Credit Loss (ECL) methodologies in many cases, some jurisdictions and internal practices still reference concepts from the incurred loss model. Understanding the incurred loss model helps risk committees and management interpret legacy outputs, reconcile comparative periods, and defend prior provisioning decisions to auditors and regulators.

Credit officers, CFOs, Model Validation teams and Risk Committees need to know when legacy incurred‑loss logic remains relevant (for example, in carve‑outs or in reconciliations) and how to translate those results into an IFRS 9 framework that uses PD, LGD and EAD Models. When stakeholders ask for an ECL comparison between methods, you’ll need clear, repeatable explanations based on data and controls.

2. The incurred loss model: core concept, components and examples

Definition

The incurred loss model recognises credit losses when a loss event has already occurred and can be identified — for example a default, significant deterioration or a past loss event. Unlike an expected loss framework, incurred models do not systematically incorporate probability‑weighted, forward‑looking scenarios into provisioning before observable loss evidence emerges.

Core components and contrast to ECL methodology

• Trigger event: identification of a credit loss event (e.g., charge‑off, default, bankruptcy).
• Measurement: losses measured based on incurred events and historical recovery rates rather than lifetime probability‑weighted expectations.
• Delay and recognition timing: recognition lags until observable impairment.

Example

Example — term loan portfolio of 10,000 small business loans: Under an incurred approach, you might recognise a loss only after a borrower defaults or you observe delinquency beyond a material threshold. If historical charge‑offs average 2% annually, the incurred model might only capture a portion of expected future losses until defaults appear. Under an IFRS 9 ECL approach you would estimate expected lifetime losses now using PD, LGD and EAD Models and forward economic scenarios. See a short primer in our Introduction to ECL for foundational concepts.

Relation to PD, LGD and EAD Models

Traditional incurred frameworks may not require explicit PD, LGD and EAD Models in the same way as ECL methodologies, but mapping between them is critical for reconciliation, scenario analysis and stress testing. For internal transparency, many banks back‑cast incurred results to implied PD/LGD/EAD to compare with their forward‑looking models; learn the underlying math in our ECL formula article.

3. Practical use cases, recurring scenarios and challenges

When you still rely on incurred outputs

Common situations where incurred outputs appear in IFRS 9 environments:

• Legacy contracts and carve‑outs during mergers—comparing pre‑acquisition incurred provisioning to post‑acquisition ECL estimates.
• Regulatory or tax reporting that references incurred results in parallel with IFRS 9 numbers.
• Short‑term credit monitoring dashboards that show observed charge‑offs versus expected provisioning.

Realistic scenario — reconciliations for the Risk Committee

Scenario: At quarter end, the Risk Committee receives two sets of figures: incurred charge‑offs of 15 bps and an ECL provision estimate of 40 bps. The committee asks why the forward‑looking model shows materially higher losses. A practical reconciliation would: (1) attribute the gap to lifetime PD assumptions and macro scenarios, (2) show historic recognition lags under the incurred approach, and (3) present sensitivity tables and a staged bridge using Three‑Stage Classification logic.

Data needs and Historical Data and Calibration

Practical calibration often requires long, clean loss histories. If your historical dataset is limited or biased, calibrations will be fragile. For modelers, combining internal loss history with external benchmarks is typical; our guidance on ECL data explains sources, augmentation and quality controls.

4. Impact on decisions, performance and stakeholder outcomes

Which areas are affected by relying on or transitioning away from the incurred loss model?

• Profitability reporting — delayed recognition in an incurred approach can temporarily overstate earnings relative to an ECL approach that provisions earlier.
• Capital planning — capital buffers and ICAAP may be understated if expected losses are not fully reflected in provisions.
• Risk appetite and pricing — product pricing that ignores lifetime expected losses can misprice risk and erode returns.
• Audit and regulatory comfort — delayed recognition may raise concerns from auditors; be ready to explain controls, staging and model governance.

For executive teams, a practical benefit of understanding incurred logic is better communication to non‑technical stakeholders — for example, explaining why capital dipped when moving from an incurred to an expected‑loss basis.

5. Common mistakes with the incurred loss model and how to avoid them

1. Treating incurred outputs as equivalent to ECL

Problem: Managers assume historical charge‑offs equals expected lifetime losses. Fix: Always produce reconciliation tables and disclose assumptions; reference our guidance on ECL disclosure to align stakeholder communications.

2. Poor calibration and over‑reliance on short histories

Problem: Small sample sizes create unstable loss rate estimates. Fix: Use long‑run data where possible, supplement with external data, and document calibration choices. See the related discussion on ECL model issues for typical pitfalls during calibration.

3. Weak governance and lack of Model Validation

Problem: Models driven by judgement without independent challenge can embed bias. Fix: Implement a Model Validation program and formal review cycles; see a practical approach in our ECL model audit resource.

4. Not applying Three‑Stage Classification consistently

Problem: Misclassification between Stage 1, Stage 2 and Stage 3 leads to wrong timing of provisions. Fix: Develop clear, measurable triggers for stage movement and test them on historical portfolios with back‑testing.

6. Practical, actionable tips and a checklist

Use this checklist when you must explain, reconcile or transition from incurred loss outputs to IFRS 9 ECL frameworks.

1. Inventory: List all reports and stakeholders that still use incurred figures (Risk Committee reports, tax, statutory, etc.).
2. Data readiness: Validate your Historical Data and Calibration process — aim for at least one full credit cycle where possible.
3. Reconciliation templates: Build standard templates that map incurred events to implied PD/LGD/EAD assumptions and show the ECL comparison clearly.
4. Governance: Require Model Validation sign‑off and maintain versioned model documentation for every change.
5. Disclosure: Update management and external disclosures to explain differences—link to your existing Expected credit losses (ECL material where helpful.
6. Stress tests: Run stress scenarios to show downside range and sensitivity of both incurred and expected approaches.
7. Training: Deliver short modules to the Risk Committee and Finance teams explaining PD, LGD and EAD Models and the differences in timing.

Short practical rule: when in doubt, document the judgment, the alternative approaches you considered, and the quantitative impact of each choice.

KPIs and success metrics for evaluating incurred vs expected approaches

• Provision coverage ratio (provisions / non‑performing exposure) — shows timeliness of recognition.
• Back‑test error on PD and LGD (actual losses vs predicted) — lower is better for model accuracy.
• Stage movement accuracy (frequency of correct Stage 1→2 transitions) — monitored monthly or quarterly.
• Time to detect losses (average days from first credit deterioration to recognition) — shorter is better under ECL.
• Number of post‑implementation model changes and audit findings — target: reduce year‑on‑year.
• Stakeholder query volume on Risk Committee Reports relating to provisioning — fewer ad‑hoc queries indicate clearer reporting.

FAQ

Reference pillar article

This article is part of a larger content cluster on Expected Credit Losses. For a comprehensive view that contrasts incurred‑loss and forward‑looking models and discusses the broader economic and social implications, see our pillar guide: The Ultimate Guide: Introduction to Expected Credit Losses (ECL) – the move from incurred‑loss models to forward‑looking models and the balance between company and employee interests, with economic and social insights.

For more technical deep dives on specific topics referenced in this article, you can also consult our pieces on ECL model issues, and the practical mechanics described in ECL comparison.