Implementing Basel IV — what CRR3
taught us on the ground.
Basel IV in the US, UK and GCC will not look like the Basel text. It will look like what CRR3 implementation actually produced inside European G-SIBs.
Executive Summary
- The output floor is a programme, not a parameter. Banks that treated 72.5% as a spreadsheet input missed 6–12 months of governance, data and reporting work. It requires cross-functional standing committees from day one.
- Data lineage is the largest sustained cost driver. Model redevelopment grabs the headlines, but tracing every risk parameter back to its golden source consumed 40–60% of run-rate programme budgets across CRR3 implementations.
- Supervisory divergence creates a 15–40 bps CET1 range on the same portfolio depending on jurisdiction. Banks operating across borders need a divergence matrix before programme kick-off, not after.
- The CRR3 ground truth compresses non-EU timelines by 12–18 months. US Endgame, UK Basel 3.1 and GCC implementations can import proven governance structures, data architectures and lessons learned from multi-year European delivery.
The 5 Transferable Lessons
CRR3 was not merely a European exercise. It was the first full-scale implementation of the Basel IV framework in a complex, multi-supervisor jurisdiction. Every bank outside the EU that is approaching Basel IV will face the same structural challenges — and can shortcut them with these five lessons.
Output floor behaves like a programme, not a parameter
The 72.5% output floor is not a single number you feed into a spreadsheet. In practice, it creates a parallel capital calculation that touches every risk type, every business line, and every reporting chain. Banks that treated it as a parameter change missed 6–12 months of programme work. It requires its own governance, its own data feeds, and a standing working group between risk, finance and the business lines.
Data lineage is where run-rate cost lives
The largest sustained cost in every CRR3 programme we worked on was not model redevelopment — it was data lineage. Tracing every input to every risk parameter back through staging layers, golden sources and manual overrides. The banks that solved lineage early freed their teams to focus on optimisation. Those that deferred it are still burning through budget two years later.
Standardised approaches are the spine, not the fallback
Under Basel IV, SA-CR, SA-CCR and the Fundamental Review of the Trading Book — SA all gain structural importance. The output floor makes the standardised result the binding constraint for many portfolios. Banks that under-invested in SA data quality, SA mapping accuracy and SA reporting infrastructure found their IRB optimisation work partially stranded.
Supervisory divergence is larger than Level 1 suggests
The CRR3 text is a single regulation, but its implementation varies meaningfully across the SSM, PRA, BaFin and every NCA that retains discretion. The same portfolio can attract 15–40 basis points more or less CET1 depending on the supervisory interpretation of discretionary items. Banks operating across jurisdictions need a divergence matrix from day one.
Model lifecycle needs re-engineering, not re-validation
CRR3 did not merely require model recalibration — it changed the economics of model maintenance. With the output floor reducing the marginal capital benefit of internal models, the cost-benefit analysis of running a large IRB model estate has shifted. The winning banks re-engineered their model lifecycle: retiring models that no longer pay for themselves, consolidating validation functions, and investing in model monitoring automation.
CRR3 Programme Budget Allocation (Typical G-SIB)
Where the €40–80m actually went — data lineage dominated run-rate spend.
What the CRR3 output-floor actually did to one G-SIB: a material CET1 unlock
Anonymised European G-SIB. How the output-floor bit, what re-segmentation it forced, and the three levers that moved the Day-1 number. Mandate-specific figures withheld; public-data calibration on request. Transferable to Basel III Endgame, UK Basel 3.1 and GCC implementation.
CFO Capital-Impact Case · coming soon Publishing autumn 2026.US Basel III Endgame — Cross-Jurisdiction Comparison
The Basel IV framework is being implemented differently in every major jurisdiction. The CRR3 experience provides the ground truth that makes this comparison actionable rather than theoretical.
| Dimension | EU (CRR3) | US (Endgame) | UK (Basel 3.1) | GCC |
|---|---|---|---|---|
| Timeline | 1 Jan 2025 (live) | Re-proposed 19 Mar 2026 (comments closed 18 Jun 2026); final rule and compliance dates pending | 1 Jan 2027 (PRA final rules, PS1/26) | KSA live since 1 Jan 2023; Qatar since 1 Jan 2024; UAE date not yet published |
| Output Floor Equivalent | 72.5% phased to 2030 | Single expanded risk-based stack proposed (advanced approaches eliminated); verify against final rule | 72.5% with PRA recalibration | 72.5% Basel text; local discretion |
| Supervisor Landscape | ECB SSM + 21 NCAs | Fed / OCC / FDIC trifecta | PRA single supervisor | CBUAE, SAMA, CBB, QCB (fragmented) |
| Data Lineage Maturity | Advanced | Moderate | Advanced | Early Stage |
| Key Risk | Transitional arbitrage | Political re-scoping | IRB cliff effects | Capability gap |
Update · 11 July 2026. UK implementation date corrected: the PRA moved Basel 3.1 to 1 January 2027 (delay announced January 2025; final rules confirmed in PS1/26, January 2026). An earlier version of this table showed 1 July 2026.
Update · 11 October 2026. US and GCC cells corrected. The Federal Reserve, OCC and FDIC re-proposed the Basel III Endgame on 19 March 2026 (comments closed 18 June 2026); the re-proposal contains no compliance dates and would replace the dual calculation with a single expanded risk-based stack. In the GCC, SAMA’s Basel III final reforms have applied since 1 January 2023 and the QCB’s revised Pillar 1 capital guidelines since 1 January 2024; the CBUAE has not published a go-live date. An earlier version showed “H2 2026 (est. re-proposal)” and “2027–2028 (varied)”.
Implementation Timeline Compression
US banks that map their Endgame implementation to the CRR3 ground truth can compress their programme timeline by 12–18 months. The structural lessons around output floor governance, data lineage architecture and standardised approach investment are directly transferable. The political uncertainty around the Endgame re-proposal is no reason to delay — the implementation architecture is the same regardless of where the final thresholds land.
Divergence Management Strategies by Jurisdiction
How leading banks are navigating supervisory differences across borders.
EU Multi-Entity Banks
- NCA discretion mapping across all entities
- Centralized floor monitoring with local interpretation layers
- SSM engagement strategy for ambiguous areas
- Transitional vs. permanent divergence tracking
US G-SIBs Preparing for Endgame
- Single expanded risk-based stack design (the March 2026 re-proposal eliminates the advanced approaches)
- Political scenario planning for threshold uncertainty
- Fed / OCC / FDIC alignment monitoring
- CRR3 pattern import for programme acceleration
UK Banks under PRA Basel 3.1
- PRA recalibration impact analysis
- IRB cliff-effect mitigation via SA investment
- Single-supervisor engagement advantage
- EU equivalence monitoring for cross-border impact
GCC Tier-1 Banks
- Proactive central bank engagement programme
- Capability-first approach to close readiness gap
- Concentrated balance sheet advantage for faster delivery
- European advisor import for proven implementation patterns
The GCC Challenge
GCC banks are approaching Basel IV from a fundamentally different starting position than European or US peers. Three structural challenges define the implementation landscape.
IRB Transferability Is Lower Than It Looks
European IRB models were calibrated on decades of through-the-cycle loss data. GCC portfolios — dominated by sovereign-adjacent corporates, real estate and construction — have different risk dynamics. Attempting to transplant EU-developed IRB approaches without re-calibration to local loss histories will produce models that supervisors will challenge and that capital planning cannot rely on. The output floor makes this less punishing but does not eliminate the problem.
Data Readiness Gap Is Larger but More Solvable
The data infrastructure in most GCC banks is 5–8 years behind the European G-SIB standard for regulatory risk data. But the advantage is a smaller, more concentrated balance sheet. A GCC Tier-1 bank can build a compliant data lineage in 18–24 months with the right architecture. European banks took 4–5 years because they were working across 15+ legal entities and dozens of legacy systems. The GCC path is shorter if it starts now.
Supervisory Relationship Is Different in Kind
GCC central banks operate with closer, more relationship-driven supervisory models than the ECB’s industrialised inspection framework. This creates both opportunity and risk. Banks that engage proactively — presenting implementation plans, requesting early guidance on interpretation — can shape the regulatory outcome. Those that wait for prescriptive instructions will find Basel IV arrives as a surprise, with less time to respond.
GCC Data Readiness vs. European G-SIB Benchmark
Maturity assessment across key Basel IV data dimensions.
GCC banks whose go-live is still ahead (the CBUAE has not published a date; SAMA has been live since 2023 and the QCB since 2024) have 18–24 months of effective build time remaining. Data lineage alone requires 12–18 months of sustained effort. Banks that have not started their data foundations by Q3 2026 will face a binary choice: accelerate with external support or accept a compliance-minimum implementation that leaves 30–60 bps of CET1 on the table.
The Implementation Playbook
Six phases, derived from the CRR3 ground truth, that form the architecture of a successful Basel IV implementation — regardless of jurisdiction.
- Calculate RWA under both IRB and SA for every material portfolio
- Identify portfolios where SA is already binding
- Establish the true output floor impact at entity and group level
- Build the dual-stack reporting infrastructure from day one
- Map every risk parameter input to its golden source
- Document manual overrides and exception processes
- Build automated data quality monitoring dashboards
- Establish lineage governance with clear ownership
- Create a standing output floor working group across Risk, Finance and Business
- Establish quarterly floor impact reporting to ExCo
- Build scenario analysis capability for floor sensitivity
- Integrate floor impact into capital allocation and pricing
- Audit the IRB model estate for cost-benefit under the new floor
- Retire models that no longer justify their maintenance cost
- Invest in model monitoring automation
- Consolidate validation functions where possible
- Map every supervisory discretion item in your jurisdiction
- Quantify the CET1 impact range for each discretion
- Monitor regulatory developments in peer jurisdictions
- Build proactive engagement strategy with your supervisor
- Benchmark your programme against EU implementation outcomes
- Adopt proven governance structures and programme architecture
- Import tested data lineage patterns and reporting templates
- Engage advisors with hands-on CRR3 delivery experience
Critical Success Factors
The four levers that separated successful CRR3 programmes from compliance-minimum outcomes.
Early Dual-Stack Architecture
Building IRB and SA reporting infrastructure in parallel from programme inception, rather than retrofitting SA capability when the floor became binding.
Data Lineage as Sprint Zero
Treating data lineage as the first deliverable, not a parallel workstream. Every downstream activity — model development, reporting, optimisation — depends on it.
Cross-Functional Governance
Standing working groups with Risk, Finance and Business representation that met at minimum monthly. Programmes without this structure lost 3–6 months to coordination failures.
Model Estate Rationalisation
Proactive retirement of IRB models where the output floor eliminated their capital benefit. The best programmes reduced model estate by 15–25% while maintaining regulatory compliance.
The Part That Will Matter in 2030
When the output floor is fully phased in and Basel IV is operational across the G20, the banks that performed well will not be those that read the regulation most carefully. They will be the ones that imported the ground truth from CRR3 and compressed their learning curve.
“CRR3 is not the final destination. It is the dress rehearsal. Every bank outside the EU that is approaching Basel IV has the opportunity to learn from multi-year European ground truth — and arrive at implementation having already made the mistakes that cost European G-SIBs hundreds of millions of euros.”— Hannan Mohammad, Founder & Managing Partner, Ezelman
Across every CRR3 programme we have delivered, the difference between a well-executed implementation and a compliance-minimum implementation is 30–60 basis points of CET1. That is not a rounding error — it is the difference between a bank that can grow and one that is capital-constrained. The same gap will emerge in the US Endgame, in UK Basel 3.1, and in every GCC implementation. The question is whether your bank will be on the right side of that gap.
The Ground Truth Value Chain
Related Intelligence
Primary documents cited in this analysis
- [1]
- [2]
- [3]
Dates, paragraph numbers and document versions verified at publication. Where consultations are still open, we flag draft status in text. Corrections: research@ezelman.com.
This analysis underpins our CRR3 leverage-ratio and output-floor mandates.
Partner-led, senior-only, no audit conflict. If the question in this piece is live at your institution, the first conversation is with a partner — not a BD team.
Talk to a partner →