CRR vs Basel 3.1¶

UK firms transition from CRR to Basel 3.1 (PRA PS1/26) on 1 January 2027, with the output floor phasing in through 2032. This section brings together all framework comparison content in one place.

Why Basel 3.1?¶

Basel 3.1 addresses three core problems the BCBS and PRA identified with the post-crisis CRR framework:

Excessive risk-weight variability¶

BCBS studies found a "worrying degree of variability" in how banks calculated risk-weighted assets for identical portfolios. Under CRR, IRB banks had wide latitude in estimating PD, LGD, and EAD — different modelling choices produced materially different RWAs for the same risk, undermining confidence in reported capital ratios.

Inadequate capital requirements in certain areas¶

The financial crisis exposed that some risk weights were too low. The standardised approach lacked granularity (e.g., all unrated corporates received 100% regardless of actual risk), and internal models in some asset classes produced capital requirements that did not reflect true economic risk.

Excessive complexity and model risk in IRB approaches¶

The IRB framework had grown so complex that it was difficult for supervisors to validate and for firms to implement consistently. The complexity itself became a source of risk — models were opaque, hard to compare, and in some cases allowed firms to optimise their way to lower capital without reducing actual risk.

How Basel 3.1 responds¶

Problem	Basel 3.1 Solution
RWA variability	Output floor — IRB RWAs cannot fall below 72.5% of SA-equivalent
SA too crude	More risk-sensitive standardised approaches — finer granularity by LTV, credit quality, etc.
IRB too permissive	Constraints on internal models — A-IRB removed for large corporates/banks/FIs; input floors on PD/LGD
Under-capitalisation	Recalibrated risk weights in areas the crisis showed were too low

The PRA adopted Basel 3.1 with targeted UK adjustments: more risk-sensitive treatment of unrated corporates, SME and infrastructure lending adjustments to avoid capital cliff-edges when existing support factors are removed, and alignment with international standards to maintain credibility and competitiveness.

Sources

At a Glance¶

Area	CRR (until 31 Dec 2026)	Basel 3.1 (from 1 Jan 2027)
1.06 Scaling Factor	Applied to all IRB RWA	Removed
Output Floor	None	72.5% of SA (phased from 50%)
SME Supporting Factor	0.7619 / 0.85 tiered	Removed
Infrastructure Factor	0.75	Removed
PD Floors	0.03% uniform	Differentiated (0.03%–0.10%)
A-IRB LGD Floors	None	Yes (5%–50% by collateral)
IRB Restrictions	F-IRB or A-IRB for all	Large corp/bank/financial sector: F-IRB only; equity: SA only
SA Risk Weights	Flat per CQS	LTV-based RE, revised corporate weights
Retail Risk Weights	Flat 75%	Differentiated: 45% transactor, 35% payroll/pension
Equity Risk Weights	100% (standard)	250%/400% (with transitional phase-in)
CCF (Unconditionally Cancellable)	0%	10%
Haircuts (Equities)	15% / 25%	25% / 35%
Currency Mismatch	None	1.5x RW multiplier for unhedged FX retail/RE
CRM Methods	Scattered provisions	Foundation Collateral Method, Parameter Substitution
COREP Templates	C prefix (C 07.00, C 08.01–08.07, C 09.01–09.02)	OF prefix (OF 07.00, OF 08.01–08.07, OF 09.01–09.02)

In This Section¶

Key Differences

Comprehensive side-by-side comparison of all regulatory parameters — risk weights, IRB floors, supporting factors, CCFs, slotting, and capital impact analysis.
Reporting Differences

COREP template changes — column additions and removals, expanded risk weight bands, new real estate breakdowns, output floor columns, and post-model adjustments.
Disclosure Differences

Pillar III disclosure template changes — OV1 output floor rows, CR5 expanded risk weight columns, CR6 post-model adjustments, CR7-A slotting CRM, CR10 HVCRE split.
Impact Analysis

Dual-framework comparison tooling, capital impact attribution waterfall, and transitional output floor schedule modelling.
Technical Reference

Developer-facing specification of parameter differences — PD/LGD floors, supervisory LGD, slotting weights, and configuration examples.