Overview
IFRS 9 fundamentally changed how banks recognise and measure credit losses—replacing the incurred loss model of IAS 39 with a forward-looking expected credit loss (ECL) approach. For a bank with a large and diverse credit portfolio, implementing IFRS 9 is not just an accounting change; it is a systems, data, and modelling programme that touches credit risk engines, core banking systems, GL interfaces, and external financial reporting.
IndusEuro provided business analysis and implementation support to a large European investment bank's IFRS compliance programme—focusing on the data flows, system requirements, and testing that connect the bank's credit risk models to its published financial statements.
The Challenge
IFRS 9 – The Business Impact
IFRS 9 requires banks to classify financial instruments into three stages based on credit deterioration, and to recognise credit loss allowances accordingly:
- Stage 1: Performing exposures. Allowance = 12-month ECL.
- Stage 2: Significant increase in credit risk since initial recognition. Allowance = Lifetime ECL.
- Stage 3: Credit-impaired (defaulted). Allowance = Lifetime ECL with evidence-based measurement.
The staging logic—determining when an exposure moves from Stage 1 to Stage 2 or Stage 3—requires quantitative criteria (relative and absolute PD deterioration thresholds) and qualitative backstops (forbearance, watchlist status, restructuring). These rules had to be specified precisely, implemented consistently, and auditable.
ECL Model Integration
ECL calculations require three core risk parameters: Probability of Default (PD), Loss Given Default (LGD), and Exposure at Default (EAD). These are produced by the bank's credit risk model systems, but the IFRS 9 variants of these parameters differ from the regulatory (Basel) variants in several important respects:
- IFRS 9 PDs must be point-in-time (PiT), not through-the-cycle (TtC) as used in Basel
- IFRS 9 LGD must be discounted at the original effective interest rate
- IFRS 9 ECL must incorporate forward-looking macroeconomic information (macro overlays)
Bridging the regulatory models and the IFRS 9 accounting models—while maintaining clear lineage from model inputs to balance sheet provisions—was a core challenge.
Reporting Chain Complexity
The reporting chain from ECL calculation to published financial statement involves multiple systems and handoffs:
Credit risk models (PD, LGD, EAD)
│
ECL calculation engine
│
Staging logic (Stage 1/2/3 allocation)
│
Provision amounts by exposure / segment
│
General Ledger (GL) posting
│
Consolidation system
│
Financial reporting system
│
IFRS 7 disclosures in Annual Report / Pillar 3
Each system handoff required analysis of data formats, reconciliation controls, and transformation logic. A figure entering the GL had to be traceable back to the underlying exposure and the specific ECL model run that generated it.
Our Approach
IFRS 9 Business Requirement Analysis
We produced business requirement documents covering:
Staging criteria specification:
- Quantitative staging triggers: relative (100% PD increase) and absolute (threshold PD levels) for Stage 1 to Stage 2 movement
- Qualitative backstops: forbearance, restructuring, watchlist criteria
- Transition logic: cure periods for Stage 2 to Stage 1 migration; cure and resolution for Stage 3
- Operational simplifications (low credit risk exemption for high-quality instruments)
ECL model parameter requirements:
- PiT PD term structure specification: methodology for converting TtC regulatory PDs to PiT IFRS 9 PDs
- Macro overlay specification: selection of macroeconomic scenarios (base, upside, downside), probability weights, and mechanism for incorporating into ECL calculations
- LGD discounting: specification of the discount rate calculation (original effective interest rate) by product type
- EAD: credit conversion factors for undrawn commitments
Collective vs. individual assessment rules:
- Criteria for individual assessment (significant exposures, Stage 3 with complexity)
- Segmentation framework for collective assessment (risk homogeneity groupings)
Data Mapping and Source System Analysis
For each ECL model input and output, we performed source system analysis:
- PD inputs: Sourced from the credit risk engine; format, granularity, and refresh frequency documented
- LGD inputs: Multiple LGD models by product type (corporate, retail mortgage, SME); mapping of each model to the correct exposure sub-population
- EAD inputs: Core banking system (on-balance sheet), limit management system (off-balance sheet commitments)
- Macroeconomic data: External data provider; format, update cycle, and internal governance of scenario assumptions
Data lineage documentation traced each input field from source system to ECL calculation to GL posting.
GL Interface Specification
The interface between the ECL calculation engine and the General Ledger was a critical deliverable:
- Posting logic: How ECL amounts are disaggregated into GL entries (provision charges, reversals, write-offs, recoveries) by legal entity, business unit, and currency
- Reconciliation controls: Automated reconciliation checks between the ECL system and GL totals at each processing cycle
- Re-run handling: GL adjustment logic for period-end re-runs and corrections
IFRS 7 Disclosure Analysis
IFRS 7 (Financial Instruments: Disclosures) requires extensive quantitative and qualitative disclosures about credit risk in the bank's annual report. We produced:
- Disclosure mapping: Mapping each required IFRS 7 disclosure (credit quality tables, ECL movement analysis, staging analysis, sensitivity analysis) to the system that would produce the underlying data
- Movement analysis specification: The IFRS 7 roll-forward disclosure (showing how the ECL allowance moved during the year: new originations, Stage movements, parameter changes, write-offs, FX) required a specific data structure in the ECL system to capture movements at transaction level
Testing and Quality Assurance
Model Validation Testing
We designed test scenarios to validate the ECL calculation engine:
- Unit tests: Individual staging criteria, individual ECL formula components
- Integration tests: End-to-end ECL calculation for representative portfolios across product types and credit quality bands
- Boundary tests: Staging boundary conditions (exposures near the Stage 1/2 and Stage 2/3 thresholds)
- Macro overlay tests: Verification that forward-looking macroeconomic adjustments are correctly applied
Reconciliation Testing
Critical reconciliation tests verified the integrity of the reporting chain:
- ECL system to GL: Provision amounts in ECL system reconcile to GL postings
- GL to consolidation: Consolidated amounts match sum-of-legal-entity GL amounts
- Period-end controls: Automated reconciliation checks catch data integrity issues before close
Regression Testing
When model parameters are updated (e.g., quarterly macro scenario refresh) or staging rules are amended, regression tests verify that the changes produce the expected delta in ECL amounts without unintended side effects.
Key Risk Parameters
ECL CALCULATION FRAMEWORK
Input Parameters
├── PD (Probability of Default)
│ ├── Point-in-time PD by tenor (1Y, 2Y, ... lifetime)
│ ├── Macroeconomic scenario adjustment
│ └── Cure probability (Stage 3)
│
├── LGD (Loss Given Default)
│ ├── Collateral value and type
│ ├── Discounted recovery cash flows
│ └── Cure rate for Stage 3
│
└── EAD (Exposure at Default)
├── Outstanding balance (on-balance sheet)
├── Undrawn commitment × CCF (off-balance sheet)
└── Accrued interest
Calculation Engine
├── Stage 1: ECL = PD(12M) × LGD × EAD
├── Stage 2/3: ECL = Σ(t=1 to lifetime) [PD(t) × LGD(t) × EAD(t) × DF(t)]
└── Macro weighting: ECL = Σ(scenario) [weight × ECL(scenario)]
Outputs
├── ECL allowance by exposure
├── Staging classification
├── ECL movement analysis (roll-forward)
└── GL posting entries
Outcomes
- IFRS 9 staging criteria specified and implemented consistently across the bank's credit portfolio, with documented rationale for each design decision.
- ECL model parameter requirements bridged the gap between regulatory credit risk models and IFRS 9 accounting requirements.
- Data lineage documented from source system to GL for all material ECL inputs and outputs.
- IFRS 7 disclosure requirements mapped to system data sources, with movement analysis data structures specified.
- Testing framework delivered covering unit, integration, and reconciliation test scenarios.
Technology Stack
- ECL calculation: Oracle (credit risk and IFRS engine)
- Data analysis: SQL, Python (model validation and reconciliation scripts)
- Reporting: XBRL (regulatory submissions), Excel-based disclosure templates
- Documentation: Confluence, structured mapping templates
- Test management: JIRA, Excel-based test case repository
Compliance Coverage
- IFRS 9: Financial Instruments—classification, measurement, impairment
- IFRS 7: Financial Instruments: Disclosures—credit risk, ECL, sensitivity analysis
- Basel III / CRR: Alignment between IFRS 9 provisions and regulatory capital (prudential floor, deferred tax)
- EBA guidelines: EBA/GL/2017/06 on IFRS 9 supervisory reporting expectations
Technology stack
Services delivered
Discuss a similar project
If this case study is relevant to a challenge you're facing, we'd welcome the conversation.
Book a consultation