Discover How ECL & COVID-19 Impacted Global Institutions

Why this topic matters for institutions applying IFRS 9

COVID‑19 produced rapid, correlated shocks across sectors and geographies that exposed weaknesses in assumptions used for ECL. For financial institutions and corporates this meant immediate questions: are PDs still valid? Do LGDs change when courts freeze enforcement? How to weight multiple macro scenarios? The decisions you made affected accounting results, capital management and stakeholder trust.

During the pandemic, many institutions revisited governance and documentation to demonstrate compliance and reasoned judgment under IFRS 9. If you need to explain changes in model outputs to auditors, boards, or investors, the lessons from COVID‑19 remain directly applicable to future systemic events and economic surprises such as those described in ECL during financial crises.

Core concepts: ECL methodology, components and COVID‑19 specific adjustments

Definition and IFRS 9 framework recap

IFRS 9 requires forward-looking expected credit loss provision that reflects: probability of default (PD), loss given default (LGD) and exposure at default (EAD) across a lifetime where necessary. The model output is shaped by staging (12‑month vs lifetime), macroeconomic scenarios, and management overlays. COVID‑19 forced frequent revisions across all these inputs.

How pandemic shocks altered each model component

• PD: Rapid increases in delinquency and unemployment required re‑calibration of PD curves and the addition of pandemic-specific vintages. Many institutions used short‑term uplift factors (e.g., +150–400 bps for consumer unsecured PDs during peak lockdowns) and then phased them down in scenarios.
• LGD: Recoveries slowed due to court closures and market value declines. Institutions increased LGD estimates for collateralized exposures by 10–30% where haircuts widened and liquidation times rose.
• EAD: For revolving facilities, utilisation spiked in some segments; modelling utilisiation curves required scenario-driven shocks (e.g., +20% util for SMEs during shock scenario) and covenant monitoring adjustments.

Scenario design and weighting

IFRS 9 expects probability-weighted scenarios. Many practitioners implemented three scenarios: base, downside, severe downside. COVID‑19 highlighted the need for flexible scenario timing (sharp drop then recovery) rather than simple linear declines. Choosing weights required transparent rationale—surveying market consensus, central bank forecasts, and internal stress tests.

Data, overlays and expert judgment

Where historical data lacked precedent (e.g., mass forbearance), institutions combined model outputs with management overlays and backstops. A structured process captured why an overlay was applied, its quantitative basis, and its planned reassessment frequency. For guidance on minimum inputs, see our discussion of data requirements for ECL.

Practical use cases and institutional scenarios

Below are common, realistic scenarios and how institutions responded in practice.

1. SME lending with government support

Situation: Government grants and loan guarantees reduced expected losses temporarily. Action: Separate modelling of guaranteed portion (lower LGD) and unguaranteed portion, with scenario-switch logic that de-weights guarantees if support expires. Documentation explained treatment to auditors and the risk committee.

2. Mortgage book with moratoria

Situation: Widespread payment holidays masked arrears. Action: Implement vintage analysis to track post-moratorium roll rates and use macro overlays to reflect longer-term unemployment risks. Ensure presentation aligns with guidance on presenting ECL in statements.

3. Corporate lending with sectoral shocks

Situation: Certain sectors (e.g., hospitality, aviation) experienced steep revenue declines. Action: Employ sector-specific PD uplifts, higher LGD assumptions for sectors with impaired collateral realisation, and adjust EAD for covenant breaches. Coordinate changes with stress testing teams and refer to research on macroeconomic data for ECL to calibrate scenario paths.

4. Trading book counterparties

Situation: Market volatility increased replacement costs and CVA concerns. Action: Align counterparty credit risk provisioning with observed market-implied default probabilities, and reconcile modelled ECL with changes disclosed under IFRS 7; see our piece on ECL impact on disclosures for disclosure implications.

Impact on decisions, performance and outcomes

COVID‑19 decisions around ECL affected outcomes across accounting, capital, strategic and stakeholder domains.

Accounting impact on profitability

Immediate increases in ECL raised impairment charges, reducing profit before tax in the short term. For example, a mid‑sized bank with a portfolio of $10bn could see a 20–50% increase in provisioning in the 2020 outlook depending on scenario weighting, directly compressing ROE. That trade‑off required boards to balance solvency and procyclical reporting.

Capital planning and risk appetite

Higher provisions affected CET1 ratios prompting capital buffers or access to relief measures. Robust documentation allowed some banks to explain temporary provisioning spikes to regulators and investors, limiting reputational damage. For sector‑wide implications, review findings on ECL impact on banks.

Investor relations and markets

Transparent IFRS 7 and other disclosures were crucial to reduce market uncertainty; see how detailed explanations affect rating agency assessments and equity analysts in our article on ECL impact on capital markets.

Internal controls and governance

Risk committees increased meeting cadence, required model signoffs at shorter intervals, and mandated retrospective analyses. These governance changes improved timeliness of adjustments but increased operational burden.

Common mistakes institutions made during COVID‑19 and how to avoid them

• Overreliance on historical calibrations: Many models failed because they assumed past cycles would mirror COVID‑19. Remedy: Introduce stress scenario-specific uplifts and add pandemic-era vintages for recalibration.
• Poor documentation of judgment: Insufficient rationale for overlays led to audit queries. Remedy: Maintain a judgment log linking qualitative decisions to quantitative impacts and trigger dates for review.
• Ignoring data gaps: Rushing to report without validating inputs created model risk. Remedy: Map data lineage, use external proxies sensibly, and record limitations per the institution’s data governance policy. For guidance, consult detailed standards on data requirements for ECL.
• Inconsistent scenario governance: Different teams using different macro inputs produced inconsistent provisioning. Remedy: Centralize scenario generation and publish a single set of macro scenarios endorsed by CFO and CRO.
• Reactive disclosures: Late or vague disclosures increased market volatility. Remedy: Proactively prepare IFRS 7 disclosure drafts linked to scenario choices and sensitivity tables.

Practical, actionable tips and checklists

Immediate checklist for ECL adjustments during systemic events

1. Assemble a cross‑functional ECL taskforce (model risk, finance, credit, legal, IR).
2. Document the economic rationale for each PD/LGD/EAD adjustment and the chosen macro scenarios.
3. Quantify management overlays separately and set explicit reassessment dates (e.g., quarterly).
4. Re-run backtests and stress scenarios: compare model outputs vs observed charge-offs and modify calibration where persistent divergence emerges.
5. Prepare sensitivity tables for disclosures showing the effect of ±10–30% scenario weights on provision amount.
6. Update risk committee packs with clear visual summaries, three-line reconciliations and recommended board actions.

Model calibration step-by-step for PD, LGD and EAD models

Follow this practical sequence:

1. Collect pandemic-era performance data and tag accounts with forbearance / government support flags.
2. Estimate short-term uplift factors using a combination of observed migration and stress-test outputs (e.g., apply a 1.5–3x multiplier to baseline PDs for worst months).
3. Re-estimate LGD using extended recovery curves and heightened haircuts; incorporate longer cure times.
4. Adjust EAD curves for increased utilisation and covenant-triggered drawing patterns; model behavioural responses to relief measures.
5. Run scenario-weighted ECL and produce sensitivity analysis; document assumptions for each scenario driver.
6. Hold model change and governance meetings with model risk and audit signoffs before production.

Reporting and committee advice

Produce risk committee reports that contain: executive summary, quantitative bridge (previous vs current ECL), key drivers, scenario tables, historical backtest evidence, and a clear ask (e.g., approve overlay of $Xm until next review). Consider guidance on macro themes and their application collected in discussions of economic challenges in ECL.

KPIs and success metrics

• Provision volatility (quarter-on-quarter % change) — target: explain >95% of variance with documented drivers.
• PD backtest accuracy — target: absolute deviation < 100 bps over a 12‑month horizon post-calibration.
• LGD recovery lag — monitor median recovery time vs pre‑pandemic baseline; aim to reduce variance via interventions.
• Time-to-approval for overlays — target: management decision within 10 business days of material macroshock.
• Disclosure completeness score — percentage of required IFRS 7 elements present and linked to quantitative tables.
• Audit queries closed within reporting cycle — target: 100% of high-priority model audit points closed prior to final sign-off.

FAQ

Reference pillar article

This article is part of a content cluster that expands on practical responses to systemic shocks; see the linked pillar for case-study learning: The Ultimate Guide: Why case studies are essential for understanding ECL implementation – how real‑world examples simplify complex standards.