Managing a bank's balance sheet implies addressing multiple structural risks, with Interest Rate Risk in the Banking Book (IRRBB) being one of the most significant due to its direct impact on profitability and solvency. In this article, we will discuss the two main approaches to measuring this risk, followed by an exploration of how to project Net Interest Income (NII) and the main challenges when forecasting it, with a focus on the most widely used practices in the industry.
Audio Article
The Basel regulatory framework defines two complementary approaches for measuring IRRBB. One is focused on the income statement, and the other on the economic value of the balance sheet. Both assess the entity’s sensitivity to interest rate movements but with different time horizons and objectives.
We will briefly explain both approaches to measuring IRRBB, focusing on the Net Interest Income approach.
Net Interest Income (NII) is the difference between the interest income from the institution’s assets and the interest expense from its liabilities over a specified time horizon.
In an environment of volatile interest rates and increasing regulatory pressure, accurate estimation of NII has become a strategic tool for ALM teams.
What is needed to estimate a financial institution’s NII correctly?
As discussed, financial margin involves projecting interest income and expense over a specific period. Accurate estimation depends on a range of assumptions and parameters; different assumptions will yield different figures and interpretations.
For this article, we'll simplify NII forecasting assumptions into three major components:
One of the most critical elements in estimating a company’s NII is defining the time horizon. Whether the projection is for 6 months, 1 year, 2 years, or 3 years, assumptions must be made about new business during that period. NII is derived from projecting future asset and liability volumes over the specified horizon.
Although the 12-month NII projection remains the main regulatory and internal focus, many institutions are aligning with Basel’s recommendations by adopting multi-horizon and overlapping projections. These allow analysis of interest rate impacts over various timeframes (6 months, 1 year, 2 years, etc.), improving the ability to anticipate inflection points in financial margin and supporting stronger strategic decision-making.
Mirai ALM & Liquidity supports this multi-horizon approach natively, running simultaneous NII projections across overlapping time windows under multiple rate scenarios, with a single consistent data model ensuring that each horizon draws from the same balance sheet assumptions and behavioral parameters.
Behavioral models are essential for accurately forecasting the future behavior of certain financial products with no clear contractual maturity or that permit early termination. Key examples include demand deposits, savings accounts, and loans with prepayment options. These models simulate customer reactions to changes in interest rates, market conditions, or commercial incentives, something critical for correctly projecting expected cash flows.
For instance, non-maturity deposits use models to estimate effective average life and sensitivity to offered remuneration. Mortgage loans, in turn, rely on prepayment models influenced by rate changes, refinancing campaigns, or regulatory shifts. Proper calibration of these models is vital to avoid significant deviations in margin estimates, especially under stress or high-volatility scenarios.
Mirai ALM & Liquidity includes built-in behavioral modeling for non-maturity deposits and loan prepayments, with configurable calibration parameters that can be updated as market conditions evolve, and Mirai AI extends this further by applying machine learning to improve behavioral estimates from historical contract-level data.
While institutions know what’s currently on their balance sheet, they cannot precisely predict what new business will be generated in the future. As time progresses and transactions are completed, new operations arise: this is referred to as “New Business”, and it plays a key role in NII estimation.
For example, if a bank projects over a two-year horizon, it must estimate how its balance sheet will evolve in terms of volume, products, and conditions over that period. The central question becomes: how much new business will be generated, and under what terms?
This is arguably one of the most critical components since the projected NII will vary significantly based on the volume and type of projected business. There are two main approaches for estimating new business:
The current balance sheet size is maintained, and only maturing or amortized positions are replenished. This approach avoids altering the institution’s current situation and is commonly used in regulatory reporting required by supervisors.
The balance sheet is adjusted based on projected growth or decline in asset and liability items. Institutions typically establish business growth budgets for retail segments (based on historical trends and market insights) and also plan wholesale funding and asset purchase strategies (terms, amounts, etc.).
With recent advances in computing and data analytics, some institutions are starting to adopt machine learning-based predictive models to estimate new business. These models incorporate macroeconomic variables, customer behavior patterns, and commercial campaigns, allowing more granular and adaptive balance projections. Especially in volatile or evolving regulatory environments, this approach enhances forecast accuracy. Furthermore, it’s essential to align new business projections with the institution’s risk appetite and strategic goals, an area that is receiving increasing attention from regulators.
The evolution of balance sheet volumes, along with the application of sophisticated models, determines how much new business must be originated to offset maturing positions and forecasted operations. It also affects the final interest rate of assets, liabilities, and off-balance sheet items. These estimates directly impact NII simulations.
Ultimately, accurately estimating Net Interest Income is not just about forecasting financial performance. It becomes a critical tool for evaluating the resilience of the business model under adverse scenarios. Margin sensitivity, combined with a strategic view of the balance sheet, enables institutions to protect their capacity for organic capital generation, an aspect increasingly valued by both regulators and investors.
In upcoming articles, we will discuss adverse scenarios and margin sensitivity – the ultimate goal of NII management in ALM. This requires estimating interest rate shocks and their implications on new business and behavioral models, and therefore, on margin sensitivity.
Mirai ALM & Liquidity automates NII projection workflows end to end: from behavioral model calibration and multi-horizon scenario analysis, to new business assumptions and IRRBB-compliant reporting, all connected through a single data model so that every projection is internally consistent and audit-ready.
Book a Demo →
Behavioral modeling, new business assumptions, and multi-horizon projections are evolving fast. This whitepaper explores how AI and Machine Learning are transforming the way banks forecast, simulate, and manage their balance sheet, with practical frameworks and real-world use cases for ALM and treasury teams. Download the Whitepaper →
What is the difference between the NII approach and the EVE approach to measuring IRRBB?
Both approaches measure a bank's exposure to interest rate risk but from different perspectives and time horizons. The NII approach evaluates the impact of rate movements on projected net interest income over a short to medium-term horizon, typically 12 months, making it the primary tool for understanding profitability trends under different rate scenarios. The EVE approach calculates the net present value of all future balance sheet cash flows discounted at market rates, offering a long-term structural view of economic solvency under permanent rate shocks. Regulators and ALM teams use both in combination because a bank can appear resilient in the short term while carrying significant long-term valuation risk, or vice versa.
Why is the choice of time horizon so important in NII projection?
The time horizon determines what assumptions must be made about new business, behavioral dynamics, and rate evolution over the projection period. A 6-month horizon requires fewer assumptions and yields more reliable estimates, but it misses medium-term inflection points in financial margins. A 2- or 3-year horizon captures more of the repricing cycle but amplifies the projection's sensitivity to new business assumptions and behavioral model accuracy. Most institutions now run multi-horizon and overlapping projections simultaneously, following Basel recommendations, so that ALCO can assess NII sensitivity at different points in the rate cycle rather than relying on a single static snapshot.
What makes behavioral models so critical to accurate NII estimation?
Many of the most significant items on a bank's balance sheet, including non-maturity deposits, savings accounts, and mortgage loans with prepayment options, have no fixed contractual maturity or can be terminated early by the customer. Without behavioral models, these items cannot be projected accurately because their cash flows depend on customer reactions to rate changes, commercial incentives, and market conditions rather than on contractual terms alone. A poorly calibrated behavioral model can produce NII estimates that diverge significantly from realized outcomes, particularly under stress scenarios in which customer behavior deviates from historical norms. Proper calibration and regular recalibration as conditions evolve are therefore one of the most operationally demanding aspects of NII projection.
What is the difference between a flat scenario and a dynamic scenario for new business in NII forecasting?
A flat or static scenario assumes that the balance sheet size remains constant over the projection horizon, with maturing or amortized positions replaced by equivalent new business. This approach preserves the current structure of the balance sheet and avoids commercial assumptions that are difficult to validate, which is why supervisors often require it for regulatory reporting. A dynamic or budgeted scenario adjusts the balance sheet based on projected growth or contraction in assets and liabilities, incorporating business planning assumptions about volumes, product mix, and funding strategy. Dynamic scenarios are more realistic for internal management purposes but are also more sensitive to the quality of commercial forecasts, making their assumptions a critical input for ALCO review and challenge.