Data Requirements

Essential data needed for IFRS 9 ECL calculations

Data Requirements Overview

Robust IFRS 9 ECL modeling requires comprehensive data across multiple dimensions. The quality and granularity of data directly impacts the accuracy of your ECL calculations and your ability to satisfy regulatory and audit requirements.

Loan-Level Data

Individual loan characteristics and current status

Historical Data

Past default and loss experience

Macro Data

Economic indicators and forecasts

Loan-Level Data

Each exposure in your portfolio requires detailed loan-level information to calculate ECL components and determine staging.

Core Loan Information
Data Field Description Used For
loan_id Unique identifier for each loan Tracking, reconciliation
origination_date Date the loan was originated Lifetime calculation, vintage analysis
maturity_date Expected maturity date Remaining lifetime, EAD profile
principal_amount Original loan amount EAD calculation
current_balance Outstanding balance as of reporting date EAD calculation
interest_rate Contractual interest rate EIR, discounting
product_type Loan product category Segmentation, PD modeling
Credit Risk Information
Data Field Description Used For
current_rating Current internal credit rating PD assignment, staging
origination_rating Rating at origination SICR assessment
days_past_due Number of days payment is overdue Staging (30 DPD rebuttable presumption)
watchlist_flag Indicator if on credit watchlist Staging assessment
forbearance_flag Indicator if forbearance measures applied Staging, cure period tracking
default_flag Indicator if borrower has defaulted Stage 3 classification
Collateral & Security Information
Data Field Description Used For
collateral_type Type of collateral (property, securities, etc.) LGD calculation, haircut selection
collateral_value Current market value of collateral LGD calculation
valuation_date Date of last collateral valuation Staleness check, indexation
guarantee_amount Value of any guarantees LGD calculation

Historical Loss Data

Historical data is essential for developing and calibrating PD, LGD, and EAD models. IFRS 9 requires at least one full economic cycle of data where possible.

Data History Recommendation
  • Minimum: 5 years of historical data
  • Ideal: 10+ years covering a full economic cycle
  • Include: At least one recession period
Default History
Default Events
  • Default date
  • Default definition applied
  • Exposure at default
  • Rating at default
  • Days past due at default
Recovery Data
  • Recovery amount
  • Recovery timing
  • Recovery source (collateral, guarantee, etc.)
  • Workout costs
  • Time to resolution
Rating Migration History

Track historical rating changes to build transition matrices for PD term structure modeling:

  • Rating at each observation point
  • Date of rating changes
  • Reason for rating change
  • Annual cohort tracking

Macroeconomic Data

IFRS 9 requires forward-looking information to be incorporated into ECL calculations. This includes macroeconomic variables that drive credit risk.

Common Macroeconomic Variables
GDP Growth
Real GDP growth rate
Unemployment
Unemployment rate
House Prices
House price index changes
Interest Rates
Central bank policy rates
Inflation
Consumer price index
Stock Market
Equity index performance
Scenario Requirements

IFRS 9 requires probability-weighted scenarios. Typically, at least three scenarios are used (base, optimistic, pessimistic) with associated probabilities. Forecasts should extend over the expected life of exposures, typically 3-5 years.

Data Quality Requirements

High-quality data is fundamental to reliable ECL calculations. Key quality dimensions include:

Completeness
  • All required fields populated
  • Full history available
  • No missing critical data points
  • Coverage of all portfolio segments
Accuracy
  • Values correctly recorded
  • Regular validation checks
  • Reconciliation to source systems
  • Error detection and correction
Timeliness
  • Data current as of reporting date
  • Timely updates of key risk indicators
  • Regular refresh of collateral values
  • Up-to-date macro forecasts
Consistency
  • Consistent definitions over time
  • Aligned across systems
  • Standardized coding schemes
  • Documented data lineage

Common Data Quality Checks

Implementing automated data quality checks is crucial before running ECL calculations. Below are standard validation rules for IFRS 9 datasets:

Field Category Check Type Validation Rule / Logic Potential Impact
Exposure Negative Values EAD < 0 Underestimation of ECL; negative provisions.
Zero Balances EAD = 0 (for active loans) Inclusion of closed accounts distorts average metrics.
Outliers EAD > Mean + 3*StdDev Single large exposure skewing portfolio results.
Dates Logical Order Maturity Date < Origination Date Invalid lifetime calculations; negative remaining term.
Future Origination Origination Date > Reporting Date Inclusion of future commitments prematurely.
Risk Ratings Missing Values Rating IS NULL Inability to assign PD; typically falls back to worst-case.
Invalid Codes Value not in Master Scale (e.g., 'XYZ') Mapping failures in PD lookup tables.
Interest Rates Negative Rates EIR < 0 Incorrect discounting; increasing ECL over time.
Excessive Rates EIR > 50% (context dependent) Over-discounting future cash flows.
Staging Mismatch Stage 1 AND DPD > 30 Violation of IFRS 9 rebuttable presumption.

Handling Data Gaps

When complete data is not available, entities must apply appropriate techniques while documenting assumptions and limitations.

Common Approaches

Use external data sources (rating agency data, industry benchmarks) when internal data is insufficient. Document the proxy relationship and any adjustments made.

Apply expert judgment overlays with proper governance, documentation, and periodic review. Common for new products or unprecedented economic conditions.

Pool similar exposures to achieve statistical significance. Define clear segmentation criteria and document the pooling methodology.

For low-risk or immaterial portfolios, simplified approaches may be acceptable. Document the rationale and ensure regular review of materiality thresholds.
Introduction Next: Staging Criteria