How IRRBB Standardised Algorithms Shock the Banking Book: ΔEVE, ΔNII, Six Rate Scenarios, Non-Maturity Deposits, Prepayments and Outlier Tests

Reader question: A bank can have no obvious trading position in interest rates and still lose economic value or earnings when rates move. How does the Basel interest-rate-risk-in-the-banking-book framework turn loans, deposits, bonds, hedges and behavioural assumptions into a systematic shock test?

Interest Rate Risk in the Banking Book (IRRBB) is the risk that movements in interest rates change the economic value of banking-book positions or the bank’s future net interest income. Basel measures the problem through complementary lenses rather than one universal number:

• Economic Value of Equity (EVE): a present-value measure of how the economic value of assets, liabilities and eligible off-balance-sheet cash flows changes under rate shocks;
• Net Interest Income (NII): an earnings-oriented measure of how interest income and expense change over a defined horizon as rates reprice.

The standardised framework then applies prescribed interest-rate shock scenarios, revalues or reprojects the banking book, and examines the adverse change. Under the current Basel framework effective from 1 January 2026, supervisors use six prescribed shock scenarios for EVE analysis and at least one supervisory outlier test compares the largest adverse ΔEVE with 15% of Tier 1 capital.

What this page owns — and what it does not

This page owns the public computational transformation:

banking-book cash flows + repricing dates + behavioural assumptions + rate shocks → ΔEVE and ΔNII diagnostics.

It does not replace duration and convexity, which provide instrument-level rate sensitivity; matched-maturity funds-transfer pricing, which allocates internal funding and repricing costs; FRTB trading-book market risk; or reverse stress testing.

This is banking-book risk mathematics and regulatory education. It does not assess any particular bank and is not personalized financial advice.

Why banking-book interest-rate risk exists

A bank performs maturity and repricing transformation. It can fund long-dated fixed-rate assets with deposits that reprice quickly, hold loans with prepayment options, issue term deposits with early-redemption behaviour, or hedge only part of its exposure.

Interest rates therefore change value and income through several mechanisms:

• repricing risk: assets and liabilities reset at different times;
• yield-curve risk: different maturities move by different amounts;
• basis risk: different reference rates move imperfectly together;
• option risk: borrowers or depositors change behaviour as rates move.

A single parallel shift cannot represent all four mechanisms, which is why the framework uses multiple scenarios and behavioural assumptions.

Step 1: build the banking-book cash-flow inventory

The engine begins with contractual and behavioural cash flows for positions in scope. For each position it needs quantities such as:

• principal balance;
• coupon or reference index;
• fixed or floating status;
• contractual maturity;
• next repricing date;
• cash-flow schedule;
• embedded prepayment or withdrawal options;
• hedging instruments assigned to the banking book;
• currency.

The first major error mode is therefore not the shock formula. It is an incomplete or wrongly timed cash-flow inventory.

Contractual maturity and repricing maturity are different

A ten-year floating-rate loan that resets every three months has a ten-year legal maturity but a much shorter repricing interval.

A fixed-rate loan that matures in ten years can have no contractual repricing before maturity, while a borrower prepayment option can shorten its effective economic life.

An IRRBB system that places every exposure only by final maturity loses the timing information that creates the risk.

Step 2: establish the base interest-rate curves

Each material currency needs a consistent base term structure appropriate to the bank’s measurement framework.

Cash flows are associated with discount rates, repricing rates or both. The curve construction may rely on the broader yield-curve machinery, but IRRBB owns the banking-book shock and behavioural layer built on top of those curves.

Step 3: apply the six prescribed shock shapes

The Basel standardised framework uses six interest-rate shock scenarios for EVE analysis:

• parallel up;
• parallel down;
• steepener;
• flattener;
• short-rate up;
• short-rate down.

These are not six arbitrary forecasts. They are standardized shapes designed to expose different maturity mismatches and curve-shape vulnerabilities.

Parallel shocks test broad duration mismatch

A parallel-up shock moves rates upward across maturities under the prescribed calibration. A parallel-down shock moves them downward.

For a bank with long-duration fixed-rate assets funded by short-duration liabilities, a parallel rise can reduce asset value more than liability value.

But a bank with a different balance-sheet structure can be more vulnerable to falling rates. The algorithm should calculate both directions rather than assume one is always adverse.

Steepener and flattener test curve-shape dependence

A steepener changes short and long rates differently so the curve becomes steeper. A flattener does the opposite.

These scenarios can reveal risks hidden by a parallel shift. A portfolio may have near-zero net duration while still holding opposing short- and long-maturity exposures that respond strongly when the curve twists.

Short-rate shocks isolate the front end

The short-rate-up and short-rate-down scenarios concentrate movement toward shorter maturities.

They are particularly relevant for products whose repricing, deposit beta or short-dated hedge behaviour is concentrated near the front of the curve.

A portfolio can therefore be relatively insensitive to a long-end move and still have large earnings or value sensitivity to short-rate shocks.

The shock calibration was updated for 2026

The Basel Committee recalibrated the prescribed IRRBB shock parameters in 2024, with the revised framework effective from 1 January 2026. A current implementation must therefore use the currently effective currency-specific parameters rather than a table copied from the original 2016 standard.

This is a classic update trigger: the algorithmic form can remain unchanged while supervisory parameters change.

Step 4: calculate base EVE

A stylized economic-value calculation is:

EVE_base = PV(assets) − PV(liabilities) + PV(eligible off-balance-sheet positions).

For deterministic cash flows:

PV = Σ CF_i × D(0,T_i).

For products with behavioural or embedded-option features, the cash-flow amount or timing can also change under the scenario rather than merely being discounted at a new rate.

Step 5: calculate shocked EVE

For shock scenario s:

EVE_s = PV_s(assets) − PV_s(liabilities) + PV_s(off-balance-sheet positions).

Then:

ΔEVE_s = EVE_s − EVE_base.

A negative ΔEVE represents an economic-value loss under the scenario.

The adverse supervisory measure focuses on the largest loss across the prescribed shocks rather than averaging good and bad outcomes together.

A simple ΔEVE example

Suppose base EVE is 100.

Under the six scenarios, shocked EVE values are:

• parallel up: 82;
• parallel down: 106;
• steepener: 91;
• flattener: 95;
• short up: 88;
• short down: 103.

The ΔEVE values are −18, +6, −9, −5, −12 and +3.

The largest adverse loss is 18.

If Tier 1 capital is 150:

18 / 150 = 12%.

That stylized bank is below the Basel 15% Tier 1 supervisory outlier threshold for this EVE test.

The 15% threshold is an outlier test, not a safety guarantee

Being below 15% does not prove that the bank has no IRRBB problem. Supervisors can consider governance, model quality, earnings sensitivity, basis risk, optionality and internal limits.

Likewise, breaching an outlier test is a supervisory signal requiring attention; it is not mathematically identical to insolvency.

Step 6: measure ΔNII

NII analysis asks how interest income and expense change over an earnings horizon as positions reprice, mature, roll over or behave differently.

In simplified form:

NII = interest income − interest expense.

For scenario s:

ΔNII_s = NII_s − NII_base.

The mechanics depend on assumptions about balance-sheet evolution, reinvestment, deposit repricing, new business and product floors. A bank can have modest ΔEVE but large ΔNII if short-term repricing is heavily asymmetric.

EVE and NII can disagree without either being wrong

EVE is a long-horizon present-value lens. NII is an earnings-flow lens over a specified horizon.

A long-dated fixed-rate asset funded by a deposit that reprices immediately can create a sharp near-term NII squeeze when rates rise, while the EVE effect depends on the full discounted cash-flow structure.

Because the lenses answer different questions:

small ΔEVE does not imply small ΔNII, and small ΔNII does not imply small ΔEVE.

Non-maturity deposits are a behavioural model, not a zero-maturity liability

Current accounts and other non-maturity deposits (NMDs) have no fixed contractual maturity, but deposit balances can remain with a bank for long periods and their administered rates may reprice slowly.

IRRBB therefore requires behavioural treatment rather than assigning every NMD to overnight maturity.

The model commonly separates more stable or “core” balances from more rate-sensitive balances and assigns repricing/maturity profiles subject to supervisory constraints.

Deposit beta matters to NII

If market rates rise by 100 basis points, a bank may raise a deposit rate by less than 100 basis points, by nearly the full amount, or with a delay.

A simple beta is:

Deposit beta = change in deposit rate / change in reference market rate.

A low beta can protect NII when market rates rise, but assuming an unrealistically low beta can overstate earnings resilience.

Historical beta can also change across rate regimes, customer segments and competition levels.

NMD assumptions create model risk

Two banks with identical deposit balances can report different IRRBB if they assume different core-deposit proportions, repricing lags or average lives.

That does not automatically mean one is wrong. It means behavioural assumptions are material model inputs that require empirical support, governance and conservative boundaries.

Loan prepayment is an embedded option

A fixed-rate borrower can often prepay early. When market rates fall, refinancing incentives can increase, shortening the asset’s effective life and returning principal when reinvestment yields are lower.

Therefore the cash-flow path under a downward shock may differ from the base path:

rate shock → changed prepayment behaviour → changed cash flows → changed EVE/NII.

Holding prepayment fixed across every scenario can miss an important source of optionality.

Term-deposit early redemption creates liability optionality

Customers can sometimes break term deposits before contractual maturity, perhaps with a penalty.

If market rates rise sharply, the incentive to exit an old low-rate deposit can increase. A model that assumes every deposit stays until legal maturity can understate repricing and funding sensitivity.

Borrower and depositor options can therefore move in opposite directions.

Basis risk survives even with matched repricing dates

Suppose an asset resets every three months to one benchmark and a liability resets every three months to another. The repricing dates match exactly.

If the two benchmark rates move differently, the spread changes.

Therefore:

matched tenor ≠ zero interest-rate risk.

Basis risk is a separate mechanism from repricing-gap risk.

Currency aggregation needs care

IRRBB is measured across material currencies, but offsetting a loss in one currency against a gain in another can hide a concentration if the positions are not economically fungible or if supervisory rules limit aggregation.

The engine should preserve currency-level results before applying permitted aggregation logic.

Inputs and outputs

A robust IRRBB engine can require:

• banking-book position inventory;
• contractual cash-flow schedules;
• next repricing dates;
• reference-rate identifiers and spreads;
• base discount and projection curves by currency;
• current Basel shock parameters;
• NMD segmentation and behavioural assumptions;
• deposit beta/repricing assumptions;
• loan prepayment models;
• term-deposit early-redemption assumptions;
• hedges and off-balance-sheet positions;
• Tier 1 capital for outlier testing;
• model/version and reporting date.

Outputs can include EVE by scenario, ΔEVE by scenario, worst adverse ΔEVE, ΔEVE/Tier 1 ratio, NII and ΔNII by scenario, currency contributions, behavioural-model contributions and assumption sensitivities.

Evidence polarity: what supports confidence?

Evidence for a reliable IRRBB result includes cash flows reconciling to source balances, repricing dates matching contracts, shock parameters matching the current Basel version, NMD assumptions supported by observed persistence and repricing data, prepayment models reproducing historical responses across rate regimes, hedge cash flows included consistently and period-to-period changes explainable by balance, curve or assumption movements.

Evidence against confidence includes final maturity substituted for repricing date, old shock tables retained after the 2026 recalibration, every NMD treated as overnight or extremely long-dated without evidence, prepayments held constant under large rate shocks, gains and losses netted across currencies without permitted logic, or ΔEVE changing materially after a data-mapping update with no economic change.

Counterexample: zero duration gap can still leave curve risk

A portfolio can have asset and liability duration matched in aggregate while holding large short-end and long-end positions that offset only under a parallel shift.

A steepener or flattener can reveal the hidden exposure.

Counterexample: identical contractual deposits can behave differently

Two customer segments can hold the same product but respond differently to rate changes because of relationship depth, digital switching behavior, insurance coverage or competition.

Product code alone is not a behavioural model.

Counterexample: a hedge can improve EVE and worsen NII

A long-dated hedge may protect economic value while creating near-term cash-flow or repricing effects that reduce earnings under one scenario.

The correct evaluation depends on the objective being hedged.

Counterexample: parallel shocks can miss basis losses

If an asset is linked to one benchmark and its funding to another, moving both curves by the same amount preserves the spread. Real basis stress can change that spread.

A clean parallel-shock result therefore does not falsify basis risk.

Weak links in implementation

Repricing-date error. contractual maturity is used instead of next reset date.

Shock-version drift. outdated currency shock parameters remain in production.

NMD overfitting. behavioural lives are calibrated too precisely to one historical regime.

Deposit-beta optimism. customer rates are assumed not to reprice despite evidence of competition.

Prepayment stasis. borrower behaviour is fixed across rate scenarios.

Hedge omission. eligible swaps or options are missing from banking-book cash flows.

Basis collapse. different reference indices are mapped to one curve.

Currency leakage. positions are aggregated before material-currency diagnostics.

Sign error. liability cash flows or payer/receiver swap directions are reversed.

Diagnostics: how to test the algorithm

• cash-flow reconciliation: principal and coupon cash flows reconcile to banking-book source systems.
• repricing-boundary test: place identical positions just before and after a reset date and confirm sensitivity changes logically.
• six-scenario test: all prescribed scenarios run for every material currency using the correct 2026-effective parameters.
• EVE arithmetic test: ΔEVE equals shocked EVE minus base EVE at full internal precision.
• outlier test: worst adverse ΔEVE divided by Tier 1 reproduces the supervisory comparison.
• NMD sensitivity test: vary core balance, average life and deposit beta within governed bounds.
• prepayment test: falling-rate scenarios accelerate eligible prepayments when the chosen behavioural model says they should.
• basis test: shock two linked reference curves differently and inspect spread sensitivity.
• hedge test: remove and restore a hedge to verify the intended directional effect.
• parallel implementation test: reproduce a controlled portfolio using an independent spreadsheet or code path.

What would falsify confidence?

Confidence should be withdrawn if the six scenarios cannot be reproduced from documented shock parameters; if source cash flows do not reconcile; if NMD behavior lacks empirical support; if the model is insensitive to obviously material prepayment or basis changes; if outlier ratios use capital from a different perimeter/date; or if a second implementation produces materially different EVE from the same cash flows and curves.

Alternatives and limits

The Basel standardised framework promotes comparability but cannot represent every bank’s behavioural dynamics. Banks also use internal measurement systems, earnings simulations, stochastic rate models, duration-gap analysis and scenario-specific stress tests.

IRRBB remains a model of how rate changes map into value and income. It does not predict the future path of interest rates, depositor behavior or borrower refinancing with certainty.

How this connects to the surrounding knowledge estate

The term-structure input comes from yield-curve algorithms. Instrument sensitivity connects to duration and convexity. Behavioural funding costs connect to matched-maturity FTP. Scenario severity can be extended through reverse stress testing.

Verification and update triggers

Preserve the Basel version, shock parameter table, material-currency list, cash-flow snapshot, curve set, NMD behavioural model, deposit beta assumptions, prepayment/early-redemption models, hedge inventory and Tier 1 capital source. Revalidate after Basel recalibrations, benchmark changes, deposit-product redesign, major rate-regime shifts, prepayment-model drift, acquisition/portfolio changes or unexplained EVE/NII jumps.

Primary and high-quality references

• Basel Framework, SRP31 — Interest rate risk in the banking book, current version effective 1 January 2026.
• Basel Committee on Banking Supervision, Interest rate risk in the banking book, April 2016, the original revised standard.
• Bank for International Settlements Financial Stability Institute, Interest rate risk in the banking book — Executive summary.
• Basel Committee on Banking Supervision, Recalibration of shocks for interest rate risk in the banking book, July 2024, effective from 1 January 2026.

Educational boundary: This article explains standardized banking-book interest-rate-risk measurement. It does not predict interest rates, judge a specific bank or provide personalized financial advice.