DPM 2.1: Building the foundation for integrated statistical and supervisory reporting

A strategic shift in the EU supervisory data ecosystem

The DPM Refit project, which resulted in the DPM 2.0 standard, established a critical semantic foundation for European supervisory reporting. However, this is only the first step in a much larger data architecture story. European supervisors are steadily moving from simple document collection to active, data-centric oversight. The initial framework established a common metamodel for structuring regulatory reporting requirements across different modelling approaches. Subsequent developments, however, indicate a broader plan to build an integrated ecosystem where data is defined once, stored logically, and reused across multiple frameworks.

For financial institutions, this evolution redefines the entire reporting lifecycle. Compliance is transitioning from a cyclical, template-driven task to a continuous data governance discipline, where a governed, logically integrated data foundation must feed the entire supervisory pipeline.

Understanding the DPM metamodel 2.1

Making this integration vision possible is the DPM 2.1 metamodel release, representing the next evolution of the supervisory foundation. To anchor this evolution, the DPM Alliance has launched a public consultation on the draft DPM 2.1 standard, with a deadline for feedback of 30 September, 2026. The final version is scheduled for publication in the fourth quarter of 2026. The evolved model is not a disruptive migration but an extension and refinement of the established metamodel, driven by practical experience and the need to support broader, cross-domain use cases, such as the ECB’s Integrated Reporting Framework (IReF). DPM 2.1 introduces eight key enhancements that have been identified as important enablers for the continued convergence of the EBA and ECB reporting systems. A major driver of these enhancements was the decision to base IReF on DPM 2.0, which revealed additional metamodel requirements now being incorporated into DPM 2.1. Analysis of the specifications highlights:

• Explicit JSON taxonomies: Standardizing on native JSON schemas establishes a modernized, lightweight format for machine-to-machine metadata exchange, reducing processing overhead for complex data transfers.
• Structured versioning and historization of metadata: DPM 2.1 introduces refined versioning mechanics. Instead of managing regulatory changes as isolated, ad-hoc updates, the new model tracks changes chronologically within a unified database structure, preserving historical reporting relationships over time
• Semantic hierarchies: The metamodel explicitly separates physical reporting templates from underlying logical concepts, allowing the same data, such as a specific exposure value, to be defined once and reused across completely different reporting modules without duplication.
• Further planned enhancements: The draft DPM 2.1 specifications indicate several further refinements, including an explicit identification of frameworks for both physical and logical modelling; improved version control for item names and descriptions; an explicit definition of data domains and their hierarchical structures; and a more detailed representation of compound properties, and include standardized Regex ("regular expression") for property values.

The broader institutional framework

This technical structure gains its authority from a unified, cross-border governance system that solidifies the framework’s role as a shared European standard. 

• DPM Alliance: Established in 2024 by the EBA, ECB, and EIOPA, the Alliance provides a joint governance framework to ensure consistent application of reporting standards across banking, insurance, and monetary statistics. With the Single Resolution Board (SRB) also involved, the collaboration is pivotal in dismantling structural silos between prudential, statistical, and insurance reporting. The stated objectives of the Alliance include maintaining and upgrading the DPM standard, facilitating the exchange of regulatory metadata, and reducing duplicated modelling efforts across authorities.
• Common Data Dictionary: This initiative represents the semantic core of the integrated vision. The goal is to create a single, authoritative source for the definition of regulatory concepts, reducing ambiguity and the need for manual reconciliation between risk, finance, and compliance functions.

Impact of DPM 2.1 on regulatory reporting

By unifying institutional governance and standardizing definitions under a common dictionary, the new EBA DPM 2.1 architecture has several direct, practical implications for how financial institutions manage data:

• Consistent definitions across regulatory frameworks:  Common glossaries and clearer links between legal sources and reporting definitions will reduce inconsistent use of concepts across frameworks. Where differences are necessary, the system identifies and describes them more transparently.
• Enhanced comparability between banks’ reports: Greater alignment of regulatory concepts can reduce interpretation differences and improve the consistency and comparability of reported data.
• Improved traceability: Requirements can be traced more systematically from their legal provision to the final reporting template, data point, or XBRL artifact, making the entire process easier to validate and audit.
• Efficient change management: A connected metadata model supports a structured impact chain, improving the speed and completeness of regulatory impact analysis.
• Greater reuse across domains: A common foundation increases data reuse across prudential, statistical, resolution, and insurance reporting, avoiding duplicate modeling work where common elements exist.

Connecting legal requirements with reporting implementation

The most significant strategic leap in the DPM 2.1 architecture is the ability to make the relationships between legal texts and data requirements explicit and machine-readable. Historically, the connection existed only in the minds of regulatory experts. The new DPM structure formalizes this relationship, creating an unbroken dependency chain.

The future structure of the DPM metamodel is being designed to explicitly support this, with planned blocks for legal references, master data definitions, and reporting-obligation rules. This enables a clear, versioned, and traceable path:

Figure: The integrated regulatory traceability chain

Today, these relationships are scattered across legislation, spreadsheets, and implementation documents. A richer, integrated metadata environment makes these connections explicit, giving expert judgment a more structured and transparent foundation.

Limits and Governance

It is important to note that a richer DPM metamodel will not enable a fully automated translation of legal texts into reporting requirements. Regulatory interpretation remains dependent on legal context, supervisory objectives, and expert judgment. Similarly, AI-generated mappings or impact assessments will continue to require accountable review by human experts.

The value of this emerging architecture lies in making the relationships between legal requirements, regulatory concepts, and technical artifacts more explicit and machine-interpretable, thereby providing a stronger foundation for experts to work from.

DPM 2.1 creates a foundation for AI-supported regulatory reporting use cases

A formally structured and connected metadata environment provides a substantially stronger foundation for the use of Artificial Intelligence across the entire reporting lifecycle. AI systems struggle when relationships between legal sources, regulatory concepts, and data requirements are implicit. A governed, machine-readable metadata model provides relationships explicitly, unlocking several high-value AI use cases.

Roadmap for DPM compliance

Successfully deploying machine-readable and AI-enabled applications requires a fundamental shift in how financial institutions design internal reporting workflows - a move away from siloed templates to integrated data. 

Outlook

Adapting to a data-centric supervisory model requires a systematic, long-term approach, and financial institutions should start to:

• Assess where fragmentation and data duplication exist across risk, finance, and reporting systems. This is the critical first step to understanding the scale of the challenge and opportunity.
• Manually map a single reporting requirement from its legal source to the final data point. This practical exercise will highlight the internal complexities that a connected architecture is designed to solve.
• Shift the budget focus from short-term, tactical reporting fixes to building a robust, centralized data foundation capable of supporting AI-enabled tools and automated oversight.

The direction is towards a more connected, cross-authority, and legally traceable metadata environment that can improve consistency, reuse, and transparency throughout the entire reporting chain.