Executive summary
- IRRBB covers gap risk, basis risk and option risk in non-trading positions.
- Economic value of equity measures the present-value impact of rate changes across the full life of positions; net interest income measures the accrual impact over a short horizon.
- The Basel framework defines standardised shock scenarios and an outlier test, but internal measurement should extend beyond them.
- Behavioural modelling of non-maturity deposits, prepayments and early redemptions is the dominant source of uncertainty.
- Hedging decisions must specify which perspective is being protected, because protecting both simultaneously is generally not possible.
The three components of IRRBB
Supervisory frameworks decompose interest-rate risk in the banking book into three sources. Gap risk arises from differences in the timing of repricing between assets, liabilities and off-balance-sheet items. Basis risk arises when instruments reprice against different reference rates that do not move together. Option risk arises from explicit or embedded optionality — caps, floors, prepayment rights and withdrawal rights — whose value changes asymmetrically with rates.
Gap risk is the easiest to measure and the most frequently discussed. Basis risk is often material in institutions funding one reference rate with another, and it is systematically underestimated because it disappears in parallel-shift analysis. Option risk requires valuation methods that recognise convexity; linear duration measures will misstate it.
Economic value and earnings perspectives
The economic value of equity perspective discounts all future banking-book cash flows at market rates and measures how the present value of assets less liabilities changes under a rate shock. It captures the full remaining life of positions and therefore reflects long-dated exposure that never appears in a one-year earnings measure.
The net interest income perspective projects accrual income over a defined horizon, commonly one to three years, under alternative rate paths and business assumptions. It reflects what shareholders and analysts observe in reported results, and it captures the effect of repricing decisions and margin compression that a static value measure does not.
ΔEVE = Σ_t [ CF_t × (DF_t^shocked − DF_t^base) ]
- ΔEVE
- — change in economic value of equity under the shock
- CF_t
- — net banking-book cash flow in time bucket t
- DF_t^base
- — discount factor for bucket t under the base curve
- DF_t^shocked
- — discount factor for bucket t under the shocked curve
The two measures are complementary, not substitutable. A bank can reduce earnings sensitivity by shortening asset repricing while increasing value sensitivity, or reduce value sensitivity through long-dated swaps while introducing earnings volatility. Reporting them side by side, with the trade-off made explicit, is the minimum standard for a committee that has to choose between them.
Scenarios and the outlier test
The Basel standard specifies six shock scenarios: parallel up, parallel down, steepener, flattener, short-rate up and short-rate down. Supervisory outlier tests compare the worst economic-value decline against Tier 1 capital, with a defined threshold that triggers supervisory attention. Regional implementations, including in the European Union, add further specification and additional tests on net interest income.
- The standardised scenarios are a comparability device, not a statement of plausible market outcomes.
- Internal scenario sets should include the institution's own historically adverse movements and a reverse-stress scenario identifying what rate path would exhaust capacity.
- Currency-by-currency measurement matters for banks with material exposure in more than one currency; aggregation can conceal offsetting positions that will not offset in practice.
Behavioural assumptions and their validation
Non-maturity deposits have no contractual maturity and no contractual repricing date, yet they typically fund a large part of the balance sheet. The assumed split between a stable core and a rate-sensitive portion, and the effective duration assigned to the core, determine the measured position. Prepayment models on fixed-rate mortgages and early-redemption behaviour on term deposits have a similar effect.
- Assumptions should be estimated on data covering at least one full rate cycle where available, and the absence of such data should be disclosed as a limitation.
- Caps on assumed deposit duration are a common supervisory expectation and a reasonable internal control regardless of expectation.
- Behavioural models fall within scope of independent model validation; they are frequently among the most material models a bank operates.
- Back-testing should compare modelled repricing behaviour against observed behaviour after each significant rate move.
Practical example
A bank holds 100 units of five-year fixed-rate loans funded by 100 units of current accounts. If the deposits are modelled with a three-year effective duration, the net duration gap is approximately two years and a 200 basis point parallel rise reduces economic value by roughly four units, other things equal. If the deposits are instead modelled as immediately repricing, the gap becomes five years and the same shock reduces economic value by roughly ten units.
Meanwhile the earnings picture inverts: under the immediate-repricing assumption, funding costs rise at once while loan income is fixed, so net interest income falls in year one. Under the three-year assumption, funding costs lag and year-one income is protected. The same balance sheet, two assumption sets, and four different answers depending on which measure is quoted. This is the practical reason IRRBB governance focuses on assumptions at least as much as on results.
Limitations and caveats
- All IRRBB measures are point-in-time and static unless explicitly run with balance-sheet evolution; a run-off assumption and a constant-balance-sheet assumption produce materially different earnings results.
- Parallel shifts rarely occur; basis and curve-shape movements can dominate realised outcomes.
- Behavioural parameters estimated in one rate environment may not transfer to another, and the error is not symmetric.
- Regulatory requirements differ by jurisdiction and are periodically revised; the current official text should always be consulted.
Conclusion
IRRBB measurement is straightforward in structure and difficult in assumption. The arithmetic of discounting and accrual projection is standard; the judgement about how customers behave when rates move is not.
A bank that reports both perspectives, discloses its behavioural assumptions with sensitivity ranges, and forces an explicit choice about which perspective it is managing has a functioning framework. One that reports a single number against a single limit does not.
References
- Interest rate risk in the banking book — standards — Basel Committee on Banking Supervision, 2016
- Guidelines on IRRBB and CSRBB (EBA/GL/2022/14) — European Banking Authority, 2022
- Basel Framework — SRP31: Interest rate risk in the banking book — Bank for International Settlements
Author bio
Jonas Mohamed Osman Abdelghafour, known as Yonas Osman is an actuary, FRM and financial risk professional specialising in banking, insurance, model risk, capital modelling and quantitative risk management.