1. Context and Objective
Resolution CMN 4.966/2021 (Bacen accounting regulation) redesigns the accounting treatment of financial instruments within Cosif and is directly inspired by the main pillars of IFRS 9, the international standard for financial instruments.
The objective of this article is to explain, in a clear and practical way, what is common between the standards, where there are usually relevant differences in Brazilian implementation, and what data, systems and governance requirements typically need to be addressed for compliance and auditing.
2. In Common: The 3 Pillars of IFRS 9 That Resolution 4.966 Incorporates
2.1 Classification and Measurement
Both standards condition the accounting classification of financial assets to two tests:
- Business model: how the institution intends to generate results with the asset (hold to collect cash flows, hold and sell, or trade and manage at fair value).
- Contractual cash flows test (SPPI): whether flows are solely payments of principal and interest.
In practice, this pillar directly affects the form of measurement and the location of results, in addition to enabling or not the application of the effective interest rate.
2.2 Expected Credit Loss (ECL) Replacing Incurred Loss Model
As in IFRS 9, Resolution 4.966 replaces the reactive incurred loss model with a prospective model based on expected credit loss calculation, which implies recognizing provisions from the initial classification of the operation, as well as updating estimates periodically based on past, current and forward-looking information (including macroeconomic scenarios).
2.3 Hedge Accounting
IFRS 9 modernized hedge accounting to bring it closer to risk management. Resolution 4.966 also addresses designation and recognition of hedging relationships, which tends to impact institutions with significant portfolios of derivatives, indexed funding and interest/foreign exchange hedge structures.
3. Staging and Significant Increase in Credit Risk
IFRS 9 structures loss recognition in stages, which is the conceptual basis behind the expected loss model that Resolution 4.966 incorporates for Financial Institutions. The general logic is:
- Stage 1: assets without significant increase in credit risk since origination, establishing a provision for the next 12 months.
- Stage 2: assets with significant increase in credit risk (SICR), establishing a lifetime provision.
- Stage 3: assets with objective evidence of loss and impairment, establishing a lifetime provision and interest recognition based on net balance.
IFRS 9 includes usual presumptions, such as 30 days past due for SICR and 90 days for default, but the standard requires the institution to define criteria consistent with its portfolio, risk appetite and historical evidence.
3.1 Minimum Criteria for SICR Definition
- Quantitative criterion: compare risk at origination versus risk at reporting date (e.g., relevant PD variation or rating migration).
- Qualitative criteria: watchlist entry, restructuring, renegotiation (forbearance), financial stress signals, reputational events and sectoral deterioration.
- Backstop criterion: days past due as reclassification trigger (e.g., more than 30 days past due for stage 2 migration).
4. Expected Credit Loss Calculation
The formula ECL = PD × EAD × LGD is the best-known way to express expected loss, where PD is Probability of Default, EAD is Exposure at Default and LGD is Loss Given Default. Some additional concepts include:
- Consideration of 12-month horizon for Stage 1 or lifetime for Stages 2 and 3.
- Application of effective interest rate (EIR) or equivalent to discount flows, avoiding over-provisioning in long terms.
- EAD modeling with amortization dynamics, prepayment, revolving limits, cards and guarantees.
- LGD modeling with recoveries, collection costs and timing (recovery curve), distinguishing real guarantees, personal guarantees and mitigators.
Additionally, the institution must clearly define the concept of default (loss event), write-off, guarantee treatment and cure criteria (return from Stage 2 to Stage 1).
4.1 Macroeconomic Scenarios and Probability Weighting
In the case of IFRS 9, ECL must reflect probability-weighted results, considering multiple reasonable and supportable scenarios (base, optimistic or pessimistic). The practical challenge is integrating macroeconomic assumptions (GDP, unemployment, interest rates, exchange rates, commodities) into PD and LGD, documenting the governance of these assumptions.
5. Key Considerations
- Modification and renegotiation rules: distinguish renegotiation with concession (forbearance) from mere commercial changes, evaluating whether there is derecognition or new recognition.
- POCI (purchased or originated credit-impaired): Assets already impaired at origination or acquisition have specific treatment in IFRS 9.
- Interest recognition in Stage 3: In IFRS 9, interest income must be calculated on net value (after provision) for impaired assets.
- Write-off and derecognition policies: objective and consistent criteria, aligned with collection and legal areas.
6. System and Controls Adequacy
6.1 Minimum Required Data
- Origination: date, product, term, rate, guarantees, initial rating/score, covenants, registration information.
- Behavior: days past due, curing, restructurings, used limits, roll rates, vintages.
- Recovery: receipt flow, costs, recovery time, guarantee realization, final losses (realized LGD).
- Macroeconomics: scenarios, historical series and version of assumptions.
- Accounting: Gross balance, provision, write-offs, reversals, write-off, reclassifications and audit trail.
6.2 Calculation Engine and Audit Trail
- Monthly processing and reprocessing capability (re-run) with model and assumption versioning.
- Separation between data (input), models (parameters) and results (output) to ensure traceability.
- Automatic controls: integrity validations, reconciliations with ledger, and abnormal variation alerts.
- Explainability: provision delta decomposition (stage migration, models used, write-off criteria).
7. Governance and Risk Model: Meeting Regulator and Audit Requirements
- Formal policies: classification and measurement, SICR, default, cure, write-off, renegotiation (forbearance), guarantees.
- Risk modeling: development, independent validation, backtesting, monitoring and recalibration.
- Overlays and management judgments: when to use, how to measure, how to approve and how to reverse.
- Segregation of duties: 1st line (business), 2nd line (risk/controls) and 3rd line (internal audit).
- Committees and minutes: decisions on scenarios, SICR thresholds, model changes and extraordinary events.
8. Practical Implementation Checklist
- Portfolio and instrument mapping: products, guarantees, available data and gaps.
- Policy definition: business models, default criteria, SICR, cure, write-off and forbearance.
- Historical database constitution: delinquency, recoveries, final losses; quality and granularity adjustments.
- Expected Credit Loss (ECL) calculation engine: Stages 1/2/3, PD/EAD/LGD, macroeconomic scenarios.
- Accounting integration: chart of accounts, entries, reconciliations and explanatory notes.
- Governance: validation, documentation, versioning and controls.
9. References (Recommended Reading)
- IFRS Foundation – IFRS 9 Financial Instruments
- Resolution CMN 4.966/2021
- PwC Brasil – Materials and articles on Resolution 4.966/21 and harmonization with IFRS 9
- Deloitte / Grant Thornton / KPMG – guidance articles and implementation impacts