A playbook for interest rate risk management

(10 pages)

The recent accelerated rise in global interest rates, the fastest in decades, brought the curtain down on an extended period of cheap money but provided little clarity on the longer-term outlook. In 2024, competing forces of tepid growth, geopolitical tension, and regional conflict are creating nearly equal chances of higher-for-longer benchmark rates and rapid cuts. In the banking industry, this uncertainty presents both risks and opportunities. But in the absence of recent precedent, many institutions lack the necessary playbook to tackle the challenge.

As rates have risen from their record lows, banks have in general profited from rising net interest margins (NIMs). However, if policy makers switch swiftly into cutting mode, banks may see the opposite effect. For now, futures markets predict the start of that process toward the end of 2024. In that context, the question facing risk managers is how they can retain the benefit of higher rates while preparing for cuts and managing the potential for macroeconomic surprises.

The question facing risk managers is how they can retain the benefit of higher rates while preparing for cuts and managing the potential for macroeconomic surprises.

The volatility playing out in rates markets is reflected in bank deposit trends, with customers more actively managing their cash to make the most of shifting monetary conditions. In Europe, deposits reached 63 percent of available stable funding (ASF) in 2023, compared with 57 percent in 2021.¹ In the US, conversely, the share of deposits over total liabilities fell over a similar period as money migrated to investments such as money market funds.

In the face of accelerating deposit flows, McKinsey research shows that bank risk management and funding performance has been highly variable. Between 2021 and 2023, the best-performing US and EU banks saw interest rate expenses rise 70 percent less than at the worst-performing banks (Exhibit 1). Among the drivers were better deposit and interest rate management.

Best-performing banks incur lower-interest-rate expenses and attract more deposits.

Alongside the impacts of deposit flows, funding has come under pressure from other factors, including the steady withdrawal of pandemic-related central bank liquidity facilities. Meanwhile, innovations such as instant payments have motivated customers to make faster and larger transfers. These withdrawals can happen quickly and be fueled by social media, creating a powerful new species of risk.

In the context of a more uncertain environment, regulatory authorities are doubling down on oversight of the potential impacts of rate volatility—for example, by asking banks to mitigate the potential effects of rate normalization, increasing overall scrutiny, and demanding evidence of methodology upgrades. Among European supervisory priorities for 2024–26, banks are advised to sharpen their governance and strategic frameworks to strengthen asset and liability management (ALM) and develop new funding plans and contingency measures for short-term liquidity shocks, including evaluating the adequacy of assumptions supporting some behavioral models.² In the same vein, the Basel Committee on Banking Supervision in 2023 proposed a recalibration of shocks for interest rate risk in the banking book. Banks can achieve this by extending the time series used in model calibration from the current December 2015 standard to December 2022, bringing more volatile rate distributions into the equation.

In a recent McKinsey roundtable, 40 percent of Europe, Middle East, and Africa bank treasurers said the topic that will attract most regulatory attention in the coming period is liquidity risk, followed by capital risk and interest rate risk in the banking book (IRRBB). With these risks in mind, 34 percent of treasurers said their top priorities with respect to rate risk were enhancing models and analytics, revising pricing strategies on loans and deposits, and beefing up ALM governance and monitoring capabilities.

Most participants also expected treasury teams to get more involved in strategic planning and board engagement and to engage business units more closely to define pricing strategies and product innovation (Exhibit 2).

Most banks expect more treasury involvement in strategic processes and more interaction with business units.

In response to these dynamics, we expect to see many banks revisiting the role of the treasury function in the months ahead. For many, this will mean moving away from approaches designed for the low-rate era and toward those predicated on uncertainty. In this article, we discuss how forward-looking banks are redesigning their treasury functions to obtain deeper insights into probabilities around interest rates and their impacts on pricing, customer behavior, deposits, and liquidity.

Five steps to enhancing the treasury function

To manage volatile interest rates more effectively, leading banks are revisiting practices in the treasury function that evolved during the low-interest-rate period and may no longer be fit for purpose—or at least should be updated for the new environment. Pioneers have taken steps in five broad focus areas: steering and monitoring, risk measurement and capabilities, stress testing, bank funding, and hedging.

Build efficiency and sophistication

A precondition of effective oversight of interest rate business is to ensure decision makers have a clear view of the current state of play. Currently, the standard approach across the industry is somewhat passive, meaning it is based on static or seldom-reviewed pricing and risk management decisions, often taken by relationship managers. Models are fed with low-frequency data, and banks use static fund transfer pricing (FTP) to calculate net interest margins. Monitoring often reflects regulatory timelines rather than the desire to optimize decision making.

Forward-looking banks are tackling these challenges through a more hands-on approach to steering and monitoring, including the following measures:

dynamic reviews of FTP, reflecting microsegment behaviors and pricing strategies tied to customer lifetime value and the opportunity cost of liquidity
increased product innovation to boost funding from both corporate and retail clients
ensuring access to high-quality, frequent, and granular data, with systems equipped to send early warning signals on potential changes in customer behaviors, especially to capture early signs of liquidity shifts
use of risk limits and targets as active steering mechanisms, bolstered by links to incentives
automation of reporting and monitoring, so liquidity and other events can be scaled internally much faster, backed by real-time data where possible

Upgrade IRRBB measurement and capabilities

Leading banks are getting a grip on IRRBB risk in areas such as balance sheet management, pricing, and collateral. Many have assembled dedicated teams to help them make more effective decisions. Given the threat to deposits, some are making greater use of scenario-based frameworks, bringing together liquidity and interest rate risk management. They are using real-time data to inform funding and pricing decisions.

To ensure they consider all aspects of rate risk, leading banks employ a cascade of models, feeding the outputs into steering and stress-testing frameworks, and capturing behavioral indicators that can inform balance sheet planning and hedging activities. Some banks are employing behavioral models to forecast loan acceptance rates and credit line drawings. Best practice involves using statistical grids differentiated by type of customer, product, and process phase.

When it comes to loans, some banks are leveraging AI to predict prepayments and their impacts on balance sheets and hedging requirements. Best practice in prepayments modeling is to move away from linear models and toward machine learning algorithms such as random forests to consider nonlinear relationships (for instance, between prepayments and interest rate variation) and loan features (for example, embedded options), as well as behavioral factors. We see five key steps:

Customer segmentation. Banks can use AI to achieve granulated segmentation—for example, incorporating behavioral factors.
Prepayment behavior. Banks can quantify constant prepayments and prepayments subject to criteria including interest rate levels, prepayment penalties, age of mortgage, and borrower characteristics. Leading banks establish a parent model and leverage customer segmentation to derive dedicated prepayment functions, taking into account customer protections such as statutory payment holidays.
Interest rate scenarios. Banks can employ Monte Carlo simulations and other models to analyze a range of scenarios, including extreme and regulatory scenarios, and simulate potential prepayment behaviors for each scenario.
Hedging ratios and strategy. Decision makers should evaluate the value of mortgages under different interest scenarios and derive sensitivities to economic value and P&L. They can then select hedging instruments with the aim of neutralizing scenario impacts.
Pricing. Mortgage pricing can be adjusted based on maturity and potential prepayment behavior. Banks can use fund transfer pricing, with risks handled by a dedicated team in the treasury function.

Another important focus area is deposit decay. Many banks still prioritize moving-average approaches segmented by maturity and backed by expert judgment. A best practice would be to identify a core balance through a combined expert and statistical approach, looking at trends across customer segmentation, core balance modeling, deposit volume modeling, deposit beta and pass-through rates, and replicating portfolio/hedge strategies. This would mean leveraging AI and high-frequency data relating to transactions, to estimate each account’s non-operational liquidity, which customers may be more likely to move elsewhere (see sidebar “Case study: Deposit modeling to limit deposit erosion”). Some banks also use survival models to gauge non-linearities in deposit behaviors.

Case study: Deposit modeling to limit deposit erosion

One bank achieved an equivalent of €150 million to €200 million positive P&L impact on €30 billion of deposits by using AI techniques for repricing. The tool provided transparency on the following measures:

the amount of liquidity at risk for each client—that is, the excess liquidity the client could potentially invest or move freely to other banks
the churn probability for each client, or the probability the client would move the liquidity if the bank took no action, based on client sophistication, the quality and intensity of the client’s relationship with the bank, and the level of market competition
the customer value at risk, an estimate of future revenues that would be at risk if the client moved the liquidity elsewhere (for example, including not only the opportunity cost of funding, but also revenues from related services)

Armed with this transparency, the bank was able to formulate client-specific strategies for repricing actions and product offerings (for example, investment products and transaction banking services), optimizing both its funding sources and profitability. New capabilities to support the effort included a deposits command center, producing a real-time dashboard for monitoring, including early warning triggers, sales team mobilization, and new product offering, especially for cash-rich corporate clients.

In the context of IRRBB strategy, leading banks are keeping a close eye on both deposit beta and pass-through rates (the portion of a change in the benchmark rate that is passed on to the deposit rate). They back their judgments with views on client stickiness, which they traditionally arrive at through expert judgment and market research. A more advanced approach is to derive regime-based elasticities, capturing data from historical economic cycles.

Better modeling enables more resilience: One bank’s story

A European global bank wanted to improve its forecasting in a rising-interest-rate context. Managers decided to focus more on customer behavior. They moved away from expert-judgment buffers to AI and stochastic modeling and a more focused approach to model calibration. They also updated scenario planning based on regulatory guidelines and best-in-class approaches, such as an interest rate risk in the banking book (IRRBB) dynamic balance sheet methodology. Through these changes, the bank was able to estimate its duration gap (between assets and liabilities) more accurately and thereby reduce delta economic value of equity (EVE). As a result, the bank recorded a 70-basis-point uplift in return on equity, resulting from capital savings on interest rate risk and a direct P&L impact from reduced hedging.

Finally, risks need to be optimally matched with hedges. The recent trend is to use stochastic models to support hedging decisions, enabling banks to gauge non-linearities. Forward-looking banks increasingly integrate deposit, prepayment, and pipeline modeling directly into their hedging strategies. They also ensure model risk is closely monitored, with models recalibrated frequently to reduce reliance on expert input (see sidebar “Better modeling enables more resilience: One bank’s story”).

Improve stress testing

Several players are integrating interest rate risk, credit spread risk, liquidity risk, and funding concentration risk in both regulatory and internal stress tests. Indeed, the IRRBB, liquidity risk, and market risk (credit spread risk in the banking book, or CSRBB) highlight the trade-off between capital and liquidity regulations. In short, higher capital requirements may reduce the need for excessive liquidity, and vice versa, for a bank with stable funding—a situation that remains a challenge to current regulatory frameworks.

Stress testing to measure interest rate risk is also evolving, with some banks adopting reverse stress testing (see sidebar “Enhancing Basel's interest rate risk measures: Exploring the efficacy of reverse stress testing and VAR”).

Enhancing Basel’s interest rate risk measures: Exploring the efficacy of reverse stress testing and VAR

Research conducted by a group of bank risk managers suggests that the current supervisory outlier tests for interest rate risk in the banking book (IRRBB) may not adequately address all significant risk scenarios. Specifically, the scenarios outlined in the BCBS 368 guidelines for stress-testing economic value of equity (EVE) and net interest income (NII) may fall short in identifying substantial IRRBB risks. This oversight could make it more difficult for banks to recognize material risks of loss, especially if they have complex or unconventional portfolios.

To identify more material risks, experts are recommending a shift in approach. Instead of focusing solely on extreme and plausible scenarios, they are advised to consider all possible scenarios and integrate reverse stress testing. This would involve simulating thousands of historical and hypothetical scenarios, covering almost the entire spectrum of possible yield curves. After computing NII and EVE, attention would be directed to the scenarios that could have the most adverse impact on the bank’s balance sheet.

In alignment with this proposed methodology, Australian banks will be mandated from 2025 to calculate IRRBB capital using measures of expected shortfall rather than value at risk (VAR). The change is intended to incorporate tail risk, with the new methodology utilizing data from the past seven years, coupled with a distinct one-year stress period.

In upgrading their stress-testing frameworks and interest rate strategies, banks need to balance net interest income (NII) and economic value of equity (EVE) risks that may materialize as a function of rate volatility. On NII, banks can productively apply scenario-based yield curve analysis across regulatory, market, and bank-specific variables and weigh these in the context of overall balance sheet exposures, hedges, and factors including deposits, prepayments, and committed credit lines. Additional economic risks include basis risk, option risk, and credit spread risk, which also should be measured.

Tailor planning

Bank funding plans are often generic, periodic, and spread across different frameworks and methodologies, including funding plans, capital plans, internal capital adequacy assessment processes (ICAAP), and internal liquidity adequacy assessment processes (ILAAP). They are often designed for a range of purposes and audiences and updated only when prompted by regulatory requirements. In future, banks will need dynamic, diversified, and granular funding plans—for example, tailored to products and regions. The plans should reflect flexible and contingent funding sources, central bank policies, and the trade-off between risks and costs.

Embrace dynamic hedging strategies

In the era of low rates, hedging of interest rate risk was a less prominent activity. Banks often employed simple, static, short-term, or isolated strategies, mostly aimed at protecting P&L. Few banks paid a great deal of attention to collateral management.

Now, in a more volatile rate environment, the potential for losses is much higher, suggesting banks need more sophisticated, agile, and frequent hedging to respond to shifts in interest rates, credit spreads, and customer deposit behaviors (Exhibit 3). Indeed, in 2023, the traded volume of euro-denominated interest rate derivatives increased by 3.4 times compared with 2020, according to the International Swaps and Derivatives Association.³

Banks have been more active in interest-rate hedging.

Hedging strategies are evolving to be dynamic, horizontally integrated across the organization, and wedded to risk appetite frameworks, so banks can balance P&L priorities and reductions in tail risk. On the ground, banks will likely need to recalibrate their strategies frequently, ideally leveraging a comprehensive scenario-based approach to reflect changes in the external environment. Many, for example, have already revisited hedging to reflect higher rates, but as rates fall, they will need to assess factors such as the impact of convexity on short positions. The objective of these exercises would ideally extend beyond risk mitigation to the optimization of NII (see sidebar “Replication and hedging: The upsides of NIM optimization”).

Replication and hedging: The upsides of NIM optimization

Broadly, banks may consider four approaches to replication and hedging, each of which offers benefits that will vary according to the bank’s unique asset base.

Static replication is a widely applied and robust approach that involves derivation and adjustment of cash flows from deposit volume models for deposit rate elasticity and pass-through rates. The remainder of cash flows are replicated with bonds, interest rate swaps, or loans. Future deposit growth can be incorporated if desired.

Dynamic hedging of present value of net interest margin (NIM) treats the deposit portfolio like a structured product. Banks calculate the present value of NIM arising from deposits, enabling derivation of present value sensitivity to changes in interest rates. The method supports dynamic hedging and can take into account negative convexity.

Static NIM optimization provides the recommended trade-off between granularity and sophistication on the one hand and usability on the other, and it is our preferred approach. It involves design of the fixed-income portfolio to replicate deposit balance dynamics over a sample period. The analyst then selects the portfolio yielding the most stable margin, represented by minimization of margin standard deviation of the spread between the portfolio return and deposit rate. The approach enables NIM maximization, with the caveat that shorter tenors tend to be preferred in periods of low benchmark rates.

Dynamic NIM optimization permits banks to model future interest rates with NIM and investment strategy optimized for a future horizon. Again, NIM can be maximized, but the approach requires assumptions on volume growth, and the optimization horizon may not extend to the full rate cycle.

A key principle of best-in-class hedging strategy is that a proactive, forward-looking approach tends to work best and will enable banks to hedge more points on the yield curve. And with forward-looking scenario analysis, they should be able to anticipate risks more effectively. Consider the case of a bank that was exposed to falling interest rates and did not meet the regulatory threshold for outliers under the new IRRBB rules for changes in NII. Through analysis of potential client migrations to other products and a push to help clients make those transfers, combined with a new multi-billion-dollar derivative hedging strategy, the bank brought itself within the threshold.

Banks should not view hedging as a stand-alone activity but rather as integrated with risk management, backed by investment in talent and education to ensure teams choose the right hedges for the right situation. These may be traditional interest rate derivatives but equally could be options or swaptions to bring more flexibility to the hedging strategy. AI will be table stakes to support decision making and identify risks before they materialize. A more automated approach to data analytics will likely be required. And collateral management should be a core element of hedging frameworks, with analytics employed to forecast collateral valuations and needs, optimize liquidity reserves, and mitigate margin call risk.

Next steps: Making change happen

To effectively implement change across the activities highlighted here, best practice would be to bring together modeling capabilities under a dedicated data strategy. The target state should be comprehensive capabilities, a unified and actionable scenario-based framework, and routine use of AI techniques and behavioral data for decisions around pricing and collateral. Most likely, a talent strategy also will be required to support capability building across analytics, trading, finance, pricing, and risk management.

Banks must marshal a broad range of market data to support effective modeling. The data will include all credit lines, including both on–balance sheet and off–balance sheet items, deposit lines, fixed-income assets and liabilities, capital items, and other items on the banking book. Ideally, banks would assemble 15 to 20 years of data, which would take in the previous period of rising interest rates from 2004 to 2007. Alongside these basic resources, banks need information on historical residual balances, amortization plans, optionality, currencies, indexing, counterparty information, behavioral insights, and a full set of macro data. Some cutting-edge models incorporate about 150 different features.

Armed with comprehensive data, banks can build behavioral models (for example, prepayments, deposits) to estimate parameters and infer behavioral effects in different scenarios. They can then integrate behavioral outputs into stress-testing simulations, alongside expert-based insights. Once macroeconomic data has been inputted, banks should be able to compute delta NII and EVE for three years. Visualization tools and hedging replica analysis can help teams clarify their insights and test their hedging strategies across risk factors.

Banks that have embraced the levers discussed here have set themselves on a course to more proactive and effective interest rate risk management. Through a sharper focus on high-quality data and the use of AI and scenario-based frameworks, banks have shown they can make better decisions, upgrade their hedging capabilities, optimize the cost of funding, and ensure they stay within regulatory thresholds. In short, they will be equipped to respond faster and more flexibly as interest rates enter a new era of volatility.

Explore a career with us

Search Openings

Banking on interest rates: A playbook for the new era of volatility