08 · IRRBB — EVE, NII & the SOT

Art 84, 98 CRD · EBA GL/RTS 2022–24

Purpose & history

Two lenses that disagree on purpose

ΔEVE revalues the whole banking book under rate shocks — the long-run solvency lens, punishing duration mismatch. ΔNII asks what happens to a year's earnings — the going-concern lens, punishing repricing gaps. They pull opposite ways: hedge the duration gap to protect EVE and you fix income streams that then miss rising rates in NII; run matched short books for NII and EVE barely moves but earnings whipsaw. Managing both is the actual craft of ALM, which is why Basel 2016 kept IRRBB in Pillar 2 with standardised outlier tests rather than a capital formula: the risk is real (2023's US failures were exactly unhedged EVE plus flighty deposits), but the numbers stand on behavioural assumptions — deposit cores, prepayment — no formula should pretend to own. The EBA's 2022 package added the NII SOT and standardised fallback methodologies for banks whose internal systems fail supervisory muster.

Requirements

Measurement enginesEVE (run-off balance sheet, no margins, risk-free discounting) and NII (constant balance sheet, 1y) under the six/two prescribed scenarios, per currency, aggregated per the RTS.

Behavioural modelsNMD core/non-core split (5y average-maturity cap), prepayment and pipeline models — documented, back-tested, ALCO-owned; these assumptions ARE the number.

SOT operationsQuarterly computation, outlier notification, remediation plans ready (hedging, repricing, product design); CSRBB assessed alongside.

GovernanceLimits below the SOT thresholds, model-change control on behavioural assumptions, internal audit coverage.

Calculations

A Dutch balance sheet meets the EVE test Assets: €40bn mortgages, duration 7 · funding: deposits, behavioural duration 2.5 · T1 = €10bn
ΔEVE(+200bp) ≈ (7 − 2.5) × 40 × 2% = €3.6bn > 15%×T1 = €1.5bn → outlier
Add €25bn pay-fixed swaps (dur. 5): ΔEVE − 25×5×2% = 3.6 − 2.5 = €1.1bn → inside. That €25bn book is the "Dutch swap programme" the section note refers to.

The NII counter-pull Assets repricing ≤ 1y €10bn vs liabilities €30bn: ΔNII(+200bp) ≈ (10 − 30) × 2% = −€0.4bn vs threshold 5%×T1 = €0.5bn — inside, barely; the swaps that fixed EVE widened this gap. One balance sheet, two tests, permanent tension.

Assumption sensitivity Shorten NMD behavioural duration 2.5 → 1.5: ΔEVE jumps ≈ €40bn × 1 × 2% = +€0.8bn — one modelling committee decision moves the SOT more than most hedges. Hence the supervisory obsession with NMD evidence.

Changes & sources

Sources: Art 84/98 CRD; EBA GL on IRRBB & CSRBB (2022), RTS on the SOT scenarios and standardised methodology (2022–24).
Watch: CSRBB quantification practice and whether NII SOT breaches start driving P2R the way EVE breaches long have.

Independent analysis by Elderlake Partners, as at August 2026. Not legal, accounting or supervisory advice.