Blog

BCL AnaCredit Reporting : Compliance Manual

1. Regulatory Foundation and Reporting Perimeter

AnaCredit (Analytical Credit Datasets) serves as a critical granular credit data collection instrument within the Eurosystem. It is designed to provide the European Central Bank (ECB) and National Central Banks (NCBs) with high-fidelity data to support monetary policy analysis, macroprudential oversight, and financial stability mandates. By shifting from aggregate to instrument-level reporting, AnaCredit enables a precise assessment of credit risk and lending behavior across the Eurozone.

The regulatory framework is anchored in Regulation (EU) 2016/867 (ECB/2016/13), which is transposed in Luxembourg via Banque centrale du Luxembourg (BCL) Circular 2017/240. While the Circular facilitates national implementation, the Reporting Agent is reminded that the Regulation remains the primary binding authority in cases of divergence.

The reporting population comprises the following resident entities:

  • Credit Institutions: All institutions resident in Luxembourg, regardless of legal status.
  • Foreign Branches: Branches resident in Luxembourg whose head office is located outside the jurisdiction.
  • Legal Entities: Luxembourgish legal entities shall report the entire dataset for themselves and their foreign branches resident in other reporting Member States, except where a derogation applies.
  • The “Single Branch” Concept: In accordance with Article 1.4 of the Regulation, all Luxembourgish branches of the same foreign legal entity are considered a single branch for reporting purposes.

To mitigate the administrative burden of double reporting, the BCL may grant derogations. A derogation is applicable if, and only if, another competent NCB in a reporting Member State already collects the complete dataset for the entity or branch in question. The BCL will contact affected Reporting Agents individually to confirm these arrangements.

Upon establishing the reporting perimeter, the Reporting Agent shall proceed to identify itself and its counterparties using the following standards.

2. Counterparty Identification and Reference Data Standards

Granular counterparty identification is the fundamental requirement for data aggregation and cross-border consistency. The Reporting Agent shall ensure that every counterparty is identified with a unique, persistent identifier to maintain data lineage across the Eurosystem.

Identifier Hierarchy for Luxembourgish Counterparties

Counterparty Type Primary Identifier Source / Format
General Entities RCS Number Trade and Companies Register (Standard structure).
Public Administrations International VAT Number VAT Register (e.g., BCL is 15444328).
Investment Funds (UCI/SIF/SICAR) CSSF Matricule (Fund) 6 characters + C00000 (e.g., OxxxxxxC00000). The “LU” prefix shall not be reported.
Fund Sub-funds CSSF Matricule (Sub-fund) 6 characters + 5-character sub-fund ID (e.g., OxxxxxxCxxxxx). The “LU” prefix shall not be reported.
International Organizations RIAD Code Official ECB List. Provide Name, City, and Country as secondary attributes.

 

The Legal Entity Identifier (LEI) is mandatory whenever available. If an LEI has not been assigned, the Reporting Agent is required to report the specific protocol NEVS = 0 (“not applicable”). The provision of an LEI does not exempt the Reporting Agent from providing the National Identifier; both attributes remain mandatory for robust cross-referencing.

Critical Requirement: Internal Identifier Persistence The Counterparty Identifier (internal ID) assigned by the Reporting Agent must be unique and invariable over time. In the event of a migration to a new IT provider or core banking application, the Reporting Agent shall ensure that these identifiers remain identical to previous submissions to prevent data duplication.

For non-identifiable resident counterparties—those lacking an RCS, VAT, or CSSF number—the Reporting Agent shall contact the BCL’s Referential team to facilitate the assignment of a RIAD code.

3. Data Set Architecture: Static vs. Dynamic Variables

The AnaCredit framework is structured into ten datasets to optimize transmission efficiency by separating variables that remain constant from those that fluctuate monthly.

The Reporting Agent shall transmit data according to the following templates:

  • Monthly Template 1 (BCL_ANCRDT_T1M):
    1. Counterparty reference data (Static – BCL_ANCRDT_T1_REF)
    2. Instrument data
    3. Financial data
    4. Counterparty-instrument data
    5. Joint liabilities data
  • Monthly Template 2 (BCL_ANCRDT_T2M): 7. Protection received data 8. Instrument-protection received data 9. Counterparty risk data 10. Counterparty default data
  • Quarterly Template 2 (BCL_ANCRDT_T2Q): 6. Accounting data

Comparison: Static vs. Dynamic Variables

  • Static Variables
    • Definition: Attributes initialized during the initial transmission, assumed to remain constant throughout the instrument’s lifecycle.
    • Example (Set 2 – Instrument Data): Instrument currency, date of creation, and floor rate.
  • Dynamic Variables
    • Definition: Attributes that are reset on every monthly transmission to reflect the current state of the credit.
    • Example (Set 3 – Financial Data): Accrued interest and outstanding nominal amount (ONA).

Voluntary Reporting Requirement Reporting Agents may choose to report attributes marked as “Not required” (X) in the Regulation on a voluntary basis. However, once transmitted, this information shall be subject to the same qualitative and validation requirements as mandatory data.

Special Case: Currency Reporting Instruments denominated in CNH shall be reported using the CNY currency code. All values must be converted to Euro using ECB exchange rates. If an ECB rate is unavailable, the Reporting Agent is required to use Bloomberg exchange rates.

BCL Callout: Reporting Special Values The Reporting Agent must distinguish between “Not required” and “Non applicable”.

  • Not required: Used when the Regulation or BCL validation rules (completeness part) do not mandate the attribute.
  • Non applicable: Used only when an attribute is required but no value exists (e.g., a missing postal code).
  • Restriction: The BCL strictly limits the use of “Non applicable.” Reporting Agents shall consult the “Non applicable” sheet within the BCL validation rules file before using this value.

4. Technical Transmission Standards and Submission Channels

To ensure Eurosystem interoperability, all data shall be exchanged using the SDMX-ML (Statistical Data and Metadata eXchange) format using XML syntax.

Data transmission must occur via the following secure channels:

  • e-file: The standard regulatory transmission platform.
  • SOFIE: The specialized financial exchange application.

Each SDMX-ML file submission must include:

  1. Reporting Agent (RA): The responsible entity.
  2. Observed Agent (OA): The entity to which the credit data pertains (required for sets 2-10).
  3. Survey: The specific template (T1M, T2M, or T2Q).
  4. Reference Period: The target date.
  5. Submission Type: “Full replacement.”

The BCL enforces the “Full replacement” principle for all resubmissions; if an error is detected, the entire file for that survey/OA/period must be resubmitted. SDMX schema versioning is strictly enforced: all submissions for reference periods prior to April 2023 shall use version 1.0.7.

5. Reporting Calendars and Transmission Deadlines

Timely remittance is essential for the BCL to meet its secondary reporting obligations to the ECB.

Standard Reporting Deadlines

Frequency Data Sets Deadline
Monthly Sets 1, 2, 3, 4, 5, 7, 8, 9, 10 15th working day after the reference period.
Quarterly Set 6 (Accounting Data) 15th working day after the reference period.

The BCL “Reporting dates” calendar is the definitive authority for all submission dates. Deadlines are provisional and subject to revision by the BCL.

6. Data Quality Management and Validation Dimensions

The Reporting Agent holds primary responsibility for data quality. The BCL employs a multi-layered validation approach covering nine dimensions:

  1. Referential Integrity (RI): Mutual dependency of attributes.
  2. Consistency (CN/CS): Logic across different datasets.
  3. Completeness (CT/CY): Presence of all mandated attributes.
  4. Uniqueness (UQ/UN): Unique identification of records via key attributes.
  5. Data Compliance (TR): Header accuracy and schema usage.
  6. Plausibility (GCMA/PC): Outlier detection and portfolio evolution.
  7. External Comparison: Consistency with BSI, MIR, and FINREP reports.
  8. Referential Data Quality: Accuracy of RIAD-relevant variables (Country, LEI).
  9. Advanced Data Quality Checks: Detection of non-plausible values or excessive dummy usage.

Operational Constraints:

  • Dummy Values: The use of technically correct but factually incorrect values (e.g., 31/12/9999 for unknown dates) is strictly prohibited.
  • Error Limits: The BCL business feedbacks will display only the first 10,000 errors.

Feedback Nomenclature: Reporting Agents shall monitor the following feedback types:

  • FDBTECH: Technical acknowledgement (Success/Failure).
  • FRML-VLDR: Formal and business validation on reference data.
  • VLD-RSLT: Business validation results for credit data.
  • DQI-DTL: Detailed Data Quality Indicators.
  • DQ-RPRT: DQI Summary Report.
  • BSI-MIR: Feedback on comparisons between AnaCredit and BSI/MIR datasets.

7. Data Quality Indicators (DQI) and Compliance Thresholds

Quality is measured via the weighted Outstanding Nominal Amount (ONA) approach (Affected ONA / Total OA ONA).

Priority Classes and Qualitative Requirements

Priority Class Threshold Key Qualitative Requirements
First Class <0.75% Portfolio completeness; no missing counterparties or instruments; no referential integrity errors; no inconsistencies with Total in comparison reports (iBSI/FINREP); no missing files.
Second Class <2.5% End-user attributes; consistency (CN/CS) issues; outliers (GCMA/PC); concentration and units of scale; inconsistencies compared to Sub-Aggregates (iBSI).
Third & Fourth <3% Remaining errors. Thresholds effective from March 2025.

Alternative DQI Framework (Effective November 2025) From November 2025, the Reporting Agent shall adhere to the Alternative DQI framework, which applies a “severity factor” to the standard DQI:

  • First & Second Class: <0.75% / <2.50% (Standard and Alternative remain identical).
  • Third Class: Alternative DQI threshold <4.50%.
  • Fourth Class: Alternative DQI threshold <9.00%.

8. Error Correction, Resubmission Protocols, and Record Keeping

Corrective actions are mandatory to maintain the historical integrity of the credit database. The Reporting Agent shall adhere to the following timelines:

  • Formal Errors: Corrected within 3 working days of feedback.
  • Other Business Errors: Corrected within 5 working days. If impossible, a resubmission schedule must be provided to the BCL within this same window.
  • Comparison Discrepancies: Corrected within 20 working days. This extended window is specifically mandated by the new ECB non-compliance procedure.

Data Preservation The Reporting Agent is required to keep all AnaCredit data and related documentation for 24 months. The BCL reserves the right to request corrections for any data within this period.

BCL Contacts: