Specialised Lending¶

Specialised Lending exposures are those where repayment depends primarily on the cash flows generated by the financed assets rather than the independent capacity of the borrower. These exposures receive special treatment under both SA and IRB frameworks.

Overview¶

Specialised lending categories:

Slotting Approach¶

The Slotting Approach maps exposures to supervisory categories based on qualitative criteria, rather than estimating PD:

flowchart LR
    A[Exposure] --> B[Assess Criteria]
    B --> C{Map to Category}
    C --> D[Strong]
    C --> E[Good]
    C --> F[Satisfactory]
    C --> G[Weak]
    C --> H[Default]
    D --> I[Risk Weight]
    E --> I
    F --> I
    G --> I
    H --> J[RW = 0%]

Slotting Categories¶

Assessment Criteria¶

Each exposure is assessed against supervisory criteria:

Project Finance Criteria¶

IPRE Criteria¶

Risk Weights¶

CRR Risk Weights¶

UK CRR Art. 153(5) defines a single risk weight table (Table 1) with maturity-based splits, covering all specialised lending types (PF, OF, CF, IPRE):

Table 1 (PF, OF, CF, IPRE):

Basel 3.1 Risk Weights¶

PRA PS1/26 restructures slotting into a single Table A with subgrade columns (A/B for Strong, C/D for Good) and introduces HVCRE as a distinct sub-type with elevated weights (UK CRR has no HVCRE concept). The tables below show the default column values (B/D per Art. 153(5)(c)). Firms may optionally use the lower A/C column values for exposures with < 2.5 years residual maturity (Art. 153(5)(d)) or enhanced underwriting (Art. 153(5)(e)/(f)). See Key Differences for the full Table A with all columns.

Non-HVCRE (OF, CF, PF, IPRE):

HVCRE:

From Category to Risk Weight: the Subgrade Step¶

The SlottingCategory enum exposed by the calculator is the coarse five-bucket grade — Strong, Good, Satisfactory, Weak, Default. Under PRA PS1/26 Art. 153(5) Table A (Basel 3.1) this maps onto a finer grid where Strong splits into columns A/B and Good splits into columns C/D — Satisfactory, Weak, and Default keep a single column each. The same A/B/C/D refinement governs the Art. 158(6) Table B expected-loss rates. Practitioners do not assign A/B/C/D directly; the calculator derives the column from a small set of input fields.

Inputs you set on the loader¶

For every specialised-lending row, set:

There is no slotting_subgrade column. The A/B/C/D refinement is derived, not declared.

How the calculator picks the column¶

slotting_category = STRONG | GOOD          ──┐
is_hvcre          = True | False             │   →  Table A row + column → risk weight
is_short_maturity = True | False  (CRR only) │

• Strong, default column (B) is used unless a concession applies.
• Strong, column A is the short-maturity concession (Art. 153(5)(d), CRR Table 1 short-maturity column).
• Good, default column (D) is used unless a concession applies.
• Good, column C is the short-maturity concession.
• Satisfactory, Weak, Default — single column, is_short_maturity is ignored.

Worked example: "I have a Strong CRE exposure" → RW¶

1. Confirm the exposure is HVCRE or non-HVCRE — set is_hvcre. A standard income-producing CRE deal is non-HVCRE (is_hvcre=False); speculative development or land-bank deals meeting the HVCRE criteria are is_hvcre=True.
2. Assign slotting_category="strong" based on the qualitative slotting criteria (see Assessment Criteria above).
3. Set residual_maturity_years from the loan's contractual maturity. The calculator derives is_short_maturity = residual_maturity_years < 2.5 automatically.
4. The calculator combines (Strong × HVCRE flag × maturity) → column → risk weight.

For the actual numerical risk weights at each (category × subgrade × HVCRE) combination, do not rely on the summary tables on this page — they show only the default columns. The single source of truth is:

• Basel 3.1 Slotting spec — Subgrade Treatment (Table A columns A/B/C/D) for the full risk-weight grid (PRA PS1/26 Art. 153(5)).
• Basel 3.1 EL rates — Table B for the EL ladder (PRA PS1/26 Art. 158(6)). Note that HVCRE Table B collapses to a flat 0.4% across columns A/B/C/D for both Strong and Good — a PRA quirk that does not apply to risk weights.
• CRR Table 1 for the CRR pre-Basel-3.1 risk weights — a single table with a < 2.5yr / ≥ 2.5yr split rather than A/B/C/D columns.
• Glossary entry for a one-screen summary of how the enum maps onto the subgrade columns.

Defaulted Exposures¶

Defaulted specialised lending exposures receive a 0% risk weight (RWA = 0). Expected loss treatment is handled separately via the provisions and EL comparison framework (see Provisions).

Calculation Example¶

Exposure: - IPRE loan, £20m - LTV: 65% - DSCR: 1.25x - Prime location - Strong tenant - Category assessment: Good - Non-HVCRE, maturity >= 2.5 years

CRR Calculation:

# Category: Good, Non-HVCRE, >= 2.5yr
Risk_Weight = 90%

# RWA
RWA = EAD × Risk_Weight
RWA = £20,000,000 × 90%
RWA = £18,000,000

Basel 3.1 Calculation:

# Standard (Operational) Good = 90% (same as CRR >= 2.5yr)
RWA = £20,000,000 × 90%
RWA = £18,000,000

SA Alternative¶

Specialised lending can also be treated under SA when the slotting approach is not used:

When to use SA vs Slotting:

Implementation¶

Slotting Calculator¶

import polars as pl
from datetime import date
from rwa_calc.engine.slotting.calculator import SlottingCalculator
from rwa_calc.contracts.config import CalculationConfig

calculator = SlottingCalculator()

calculate_branch() takes a pre-filtered slotting LazyFrame and returns a LazyFrame. Build a single-row frame for a single-exposure calculation:

df = pl.DataFrame({
    "exposure_reference": ["EX1"],
    "ead": [20_000_000.0],
    "slotting_category": ["good"],
    "is_hvcre": [False],
    "is_short_maturity": [False],
    "is_pre_operational": [False],
}).lazy()

result = calculator.calculate_branch(
    df, CalculationConfig.crr(reporting_date=date(2026, 12, 31))
).collect().to_dicts()[0]
# result["rwa"] -> 18_000_000.0

Risk Weight Lookup¶

Slotting risk weights are cited entries in the rulepack packs (src/rwa_calc/rulebook/packs/{crr,b31}.py), read through the resolved pack rather than a standalone lookup table:

from datetime import date
from rwa_calc.rulebook.resolve import resolve

# Resolve the frozen, content-hashed rulepack for the regime + reporting date
pack = resolve("crr", date(2026, 12, 31))

# Slotting risk weights are cited LookupTable entries on the pack, e.g.
# the base, short-maturity, HVCRE, and HVCRE-short slotting RW tables keyed
# by slotting category (read by engine/slotting/transforms.py).
slotting_rw_base = pack.lookup("slotting_rw_base")
# Non-HVCRE, standard maturity (>= 2.5yr), category GOOD -> 0.90

The SlottingCategory enum (rwa_calc.domain.enums) is still the canonical category key used by the slotting transforms.

Project Finance Detail¶

Pre-Operational Phase¶

Project is in construction or commissioning: - Construction risk present - Cash flows not yet established - Under PRA PS1/26 slotting, pre-operational PF uses the same risk weights as operational PF - Under SA (Art. 122B), pre-operational PF receives a higher weight (130% vs 100%)

Operational Phase¶

Project is generating cash flows: - Construction complete - Revenue stream established - Standard slotting weights apply (identical to pre-operational under PRA)

Phase Transition (SA Only)¶

The pre-operational / operational distinction only affects risk weights under the SA approach (Art. 122B(2)(c)), not under IRB slotting:

# Basel 3.1 SA applies different weights for PF by project phase
if framework == "BASEL_3_1" and approach == "SA" and lending_type == "PROJECT_FINANCE":
    if phase == "pre_operational":
        rw = 1.30  # 130% (Art. 122B(2)(c))
    elif phase == "high_quality_operational":
        rw = 0.80  # 80% (Art. 122B(4))
    else:
        rw = 1.00  # 100% operational (Art. 122B(2)(c))
# For IRB slotting, all PF uses standard non-HVCRE weights regardless of phase

HVCRE Treatment¶

High Volatility Commercial Real Estate receives higher risk weights under PRA PS1/26 (Basel 3.1) due to:

• Speculative development
• No established cash flows
• Higher correlation to economic cycles

HVCRE Criteria:

• CRE development or land acquisition
• Repayment from future sale or refinancing
• Uncertain outcome

Risk Weight Comparison:

Supporting Factors¶

Infrastructure project finance may qualify for the 0.75 infrastructure supporting factor under CRR (Art. 501a). This factor is applied by the SupportingFactorCalculator after slotting risk weights are determined, not within the slotting engine itself.

See Supporting Factors for eligibility criteria and application details.

Regulatory References¶

Next Steps¶

• Credit Risk Mitigation - CRM for specialised lending
• Supporting Factors - Infrastructure factor details