Major European banks are ramping up efforts to convince regulators to allow the use of pooled credit loss data—sourced from the Global Credit Data (GCD) consortium—in their risk-weight modeling under the Basel framework and the refreshed Capital Requirements Regulation (CRR3). The push comes as the banking sector navigates the complexities of new data requirements, model comparability challenges, and heightened supervisory scrutiny expected with CRR3’s phased rollout starting January 2025.
Why Banks Want to Use GCD Data
Banks argue that while internal ratings-based (IRB) models require extensive historical data to accurately estimate parameters like Probability of Default (PD) and Loss Given Default (LGD), many have portfolios—especially low-default sectors—where historical data is inadequate. The GCD consortium, a data-sharing initiative founded in 2004 and now supported by over 50 leading banks globally, pools anonymized internal credit loss records, providing a standardized dataset with over 250,000 individual facility default records from more than 70,000 obligors across 120 countries.
By tapping into this broad dataset, banks believe they can:
-
Strengthen the statistical robustness of risk parameters for their IRB models.
-
Standardize model outputs, reducing heterogeneity in risk weights across institutions—a long-standing issue in the EU banking sector.
-
Lower model uncertainty, particularly for portfolios with “thin” internal data.
-
Mitigate procyclicality by relying on a more diversified pool of loss experiences, not just those reflecting recent trends.
The Regulatory Hurdle: Quality and Acceptance
The current CRR framework places a heavy premium on internal data quality and prohibits broad reliance on external data—bank models must be justified by their own loan books and loss histories. Banks are now urging the European Banking Authority (EBA) to revisit these rules and explicitly bless the use of vetted external datasets like GCD as a “backstop” where internal data falls short. They contend that GCD’s rigorous onboarding process aligns with Basel III definitions and delivers a level of data compatibility regulators should find acceptable.
However, authorities like the European Central Bank (ECB) and EBA have raised concerns about:
-
Governance and oversight of pooled datasets.
-
Ensuring that such data matches EU risk-type definitions and isn’t distorted by regional or sector differences.
-
Validation of anonymization and quality controls.
EBA Consultation and Next Steps
In response to industry lobbying, the EBA is consulting on whether to revise credit risk guidelines and IRB model standards to explicitly permit pooled data. This process involves scrutinizing data integrity, model governance, and estimating the impact of data blending on cross-bank comparability and systemic risk.
If the EBA and the European Commission approve the necessary amendments, banks may begin integrating GCD-based loss data into IRB model submissions, subject to further oversight and possibly transitional implementation rules. The proposed changes would also dovetail with wider efforts to harmonize risk-weight practices and account for the modernization of EU capital regulations.
Practical Implications for Banks, Risk Officers, and Compliance Teams
-
Business impact: Enhanced data sharing could ease model validation processes, lower capital charges in low-default portfolios, and support more granular risk-aware lending.
-
Audit and regulatory scrutiny: Model validation teams will need to document how pooled data is integrated, ensure alignment with EBA governance rules, and monitor for any shifts in output driven by external data inclusion.
-
Operational changes: Compliance and IT functions must prepare for enhanced data pipeline requirements, audits of pooling mechanisms, and additional documentation for model approvals.
-
Reputational aspect: By supporting industry-wide data initiatives, banks may be able to position themselves as transparent, collaborative, and attuned to best practices in risk modeling.
Meanwhile, analysts caution, “Without robust governance, pooled data can create systemic blind spots. Supervisors must ensure these datasets are truly representative, up-to-date, and free from hidden biases”.
Best Practices Going Forward
To make the most of pooled data in forthcoming IRB model reviews, banks should:
-
Develop clear governance frameworks for data validation and integration.
-
Regularly benchmark external data against internal outcomes to detect divergences.
-
Engage early and transparently with supervisors, providing detailed documentation on methodology and back-testing.
-
Educate model development and validation teams on emerging EBA expectations and transitional CRR3 rules.
FAQ
What is Global Credit Data (GCD)?
GCD is a consortium pooling anonymized credit loss data from 50+ banks, providing risk modeling support with standardized default and loss records.
Why is pooled data needed under Basel/CRR frameworks?
Many banks lack sufficient internal data, especially for low-default portfolios. Pooled data enhances model robustness and risk weight consistency.
What’s the main regulatory concern?
Authorities need assurance around data governance, alignment with EU definitions, and rigorous validation to ensure pooled datasets don’t introduce systemic weaknesses.
What is the timeline?
With CRR3 starting from January 2025, and EBA/European Commission reviews ongoing, banks may be able to use pooled GCD data in IRB models subject to successful guideline amendments and regulatory clearance.
How will this affect loan portfolios and capital requirements?
If approved, banks could better capture true underlying risk, potentially reducing capital volatility and improving access to credit in less developed risk-data segments