← All insights
Guide6 min read

The BCBS 239 principles, and why banks still fall short of them

Fourteen principles, a decade of remediation, and one cause that sits underneath most of the shortfalls.

The BCBS 239 principles are the Basel Committee's 14 principles for effective risk data aggregation and risk reporting, published in January 2013. They require a bank to produce accurate, complete and timely risk data across all its legal entities, in normal times and under stress, and to report it clearly to the board and supervisors. Global systemically important banks were expected to comply by January 2016. A decade later most still do not fully comply, and supervisors have moved from recommendations to escalation.

The 14 BCBS 239 principles

The principles fall into four groups. The first eleven apply to banks; the last three to supervisors.

Overarching governance and infrastructure 1. Governance: risk data aggregation and reporting sit within the bank's risk management framework, owned by the board and senior management. 2. Data architecture and IT infrastructure: designed and maintained to support aggregation in normal times and in stress.

Risk data aggregation capabilities 3. Accuracy and integrity: risk data is reconciled with sources, largely automated and controlled. 4. Completeness: all material risk data is captured, across business lines, legal entities and risk types. 5. Timeliness: aggregated data is available within the timeframes the board and supervisors need. 6. Adaptability: the bank can produce ad hoc aggregations quickly, including for stress and crisis.

Risk reporting practices 7. Accuracy: reports reflect risk precisely and are validated. 8. Comprehensiveness: reports cover all material risks. 9. Clarity and usefulness: reports are clear, concise and fit for their reader. 10. Frequency: set by the needs of the recipient and the nature of the risk. 11. Distribution: reports reach the right people securely and on time.

Supervisory review, tools and cooperation 12. Review: supervisors assess compliance periodically. 13. Remedial actions and supervisory measures: supervisors act on deficiencies. 14. Home and host cooperation: supervisors cooperate across jurisdictions.

What supervisors test now

The 2013 text is the base, but what a bank complies with in 2026 is layered on top of it. The Basel Committee's November 2023 progress report found that only two of the 31 global systemically important banks assessed fully complied, and that no single principle was fully implemented across all of them. The European Central Bank responded with its Guide on effective risk data aggregation and risk reporting in May 2024, which turns the principles into examinable expectations. Two of them matter most here: complete and up-to-date data lineage for critical data elements, from capture to report; and the ability to respond quickly when a supervisor asks for a specific figure, often tested through fire drills.

Enforcement has followed. The ECB has set out a staged escalation that runs from written recommendations to periodic penalty payments and sanctions, and has linked capital add-ons partly to data quality. In the United States, data management deficiencies led to civil money penalties of $135.6 million against one large bank in July 2024. The ECB's supervisory priorities for 2026 to 2028 again name risk data aggregation and reporting as a material shortcoming. In January 2026 the Basel Committee also tied the principles to AI, noting that the quality of AI-driven outputs depends on the quality of the data beneath them.

Why banks still fall short of BCBS 239

The common causes are well documented: legacy architecture, manual workarounds in end-user computing, unclear ownership of critical data, and programmes that produce documentation rather than capability. One cause sits underneath most of them.

Lineage stops where the code begins. Banks have invested in catalogues and lineage tools that trace data between databases and reporting layers. The transformations that turn source data into a risk figure often sit in application code, in Java, Python or COBOL, or in a spreadsheet maintained by one team. The lineage arrives at the edge of the application and stops. When a supervisor asks how an outcome was produced, the bank sends people into the code and the spreadsheets to reconstruct the answer, and repeats the exercise at the next request.

Checks test agreement, not correctness. Reconciliations, variance analysis and four-eyes review confirm that numbers agree with their sources and move plausibly. An error of treatment, such as a rule applied wrongly or logic not updated after a regulatory change, agrees with its sources every period and passes every check. Principle 3 asks for accuracy, not agreement.

The documented process and the operated process drift. The catalogue, the lineage tool, the data dictionary and the process documentation describe what should happen. The process as operated is what produces the figure. Nothing holds the two together between reviews, so the bank proves alignment by project, on the day, and loses it with the next change.

From BCBS 239 compliance to process context

Meeting the principles is not a documentation exercise. It needs a view of the risk reporting process that joins the documented version to the operated one, down to the transformations in the code, and that can be tested.

xflow builds that view, which we call process context. It connects the regulatory objective to the process steps and to the code, data, spreadsheets and manual routines that deliver each outcome. The code's transformations are joined into the lineage, so the lineage is executable: real data can be run through it to check whether the process produces the reported figure. Gaps such as truncated lineage, undocumented transformations and missing business terms are located, fixed and written back to the data catalogue.

At a global tier one bank, lineage for regulatory reporting stopped where the application code began. xflow connected to the code repositories and produced the first analysis in hours. Transformation gaps were closed and written back to the bank's data governance layer, giving end-to-end lineage from policy to process to code to data.

BCBS 239 is a banking standard, but the gap it exposes is not unique to it. The same gap shows up under Solvency II and IFRS 17 for insurers.

Questions about the BCBS 239 principles

What is BCBS 239?

The Basel Committee on Banking Supervision's standard number 239, Principles for effective risk data aggregation and risk reporting, published in January 2013.

Who must comply with BCBS 239?

Global systemically important banks, by January 2016. The Basel Committee recommended applying it to domestic systemically important banks within three years of designation, and many supervisors, including the ECB, apply its expectations more widely.

What does RDARR stand for?

Risk data aggregation and risk reporting, the subject of BCBS 239. The ECB uses RDARR in its 2024 Guide.

Does BCBS 239 require data lineage?

The 2013 principles imply it. The ECB's 2024 Guide makes it explicit: banks should hold complete and up-to-date lineage for their critical data elements, from where data is captured to where it is reported.

Why is BCBS 239 compliance so hard?

Because the transformations that produce risk figures sit in code, spreadsheets and manual steps that catalogues and lineage tools do not reach, and because checks confirm that numbers agree rather than that they are right.

How does BCBS 239 relate to AI?

AI systems that use risk data inherit its quality. The Basel Committee made the link in January 2026, noting that the quality of AI-driven outputs depends on high-quality data. Strong risk data aggregation is becoming a condition for using AI in risk processes.

Which outcome do you need to prove now?

Start with one critical outcome. We’ll show you how it is produced today, where that differs from what is documented, and how to close the gaps.

Start with one critical outcome