When static models stop working: navigating banking’s new normal

Banks are operating in a more volatile world than ever, but most of their interest rate risk models haven’t caught up to this reality. We explore how static assumptions and slow governance can rapidly turn manageable risk into a deeper crisis and why dynamic, adaptive modelling should be the backbone of any modern resilience strategy. 

Interest rate risk models were built for a world that barely moved. More than 13 years of low rates and predictable flows made long-run averages feel safe, but today’s environment is the opposite. Speed and volatility dominate, and confidence shocks can spread in hours. The failure of a handful of major banks in 2023 exposed the gap, because banks leaned on static assumptions and governance that couldn’t keep pace. 

This article forms part of a broader series from Baringa exploring how banks can rebalance risk and growth across the balance sheet, from the buffers they carry to the behaviour, models, and infrastructure that shape how that capacity gets used. Here, we look at the interest rate risk models underpinning banks’ liquidity strategies, and why current approaches must evolve to keep pace with the new nature of risk.  

Today’s models tend to carry three common weaknesses 

Interest rate risk in the banking book (IRRBB) measures how rate changes affect earnings and value, while traditional liquidity frameworks focus on funding stability. Both share three weaknesses: 

1. Static assumptions in a dynamic world. Behavioural parameters such as deposit stickiness, prepayment speeds, and optionality are often fixed for months or years. In a market where behaviour changes in days, these assumptions quickly become liabilities
2. Historical bias. Models anchored in a prolonged period of ultra-low rates underweight tail scenarios. The sharp hikes that began in 2022 made historical averages almost irrelevant for forecasting deposit outflows or margin compression
3. Slow recalibration. Even when assumptions are challenged, governance cycles treat updates as episodic. Behavioural dynamics now shift quarterly, sometimes weekly, leaving banks exposed while models wait for recalibration. 

These weaknesses surface across operations in the form of mispriced products, under‑hedged positions, and balance sheet strategies that assume stability where none exists. The effect is often compounded by how models are used, with limited scenario design and slow governance cycles failing to capture joint, fast‑moving stress events, leaving liquidity and IRRBB exposed together. 

Behavioural dynamics and assumptions are shifting 

Behaviour has changed structurally, and money now moves faster than ever before. Three key forces are driving this shift: 

• Online access, mobile banking and digital assets make it frictionless to move funds 

• Real‑time payments and sweep products accelerate liquidity movement 

• Social media amplifies confidence shocks at speed. 

The result is that deposits, insured or not, can leave in hours, instead of the weeks it took previously. These behavioural shifts amplify optionality. Non‑maturing deposits and embedded options behave non‑linearly under stress, with faster decay, sharper prepayment responses, and heightened liquidity strain. As banks depend on behavioural profiling to shape key metrics, hedging strategies and earnings sensitivity, this can create divergence between model‑driven insights and the actions needed to protect liquidity and resilience. 

Silicon Valley Bank’s failure showed how static deposit assumptions can mask liquidity risk until it’s too late, while the collapse of Credit Suisse highlighted the speed at which confidence can evaporate. In both cases, underlying models were not able to adapt quickly enough, leaving hedging and pricing exposed when volatility spiked. 

Dynamic modelling offers a way through the constraints of static models 

Dynamic modelling embeds flexibility and behavioural realism into the core of liquidity and IRRBB frameworks through three key approaches: 

• Continuous recalibration. Deposit decay rates, prepayment speeds, and optionality assumptions need to update as conditions evolve  

• Scenario agility. The ability to rerun scenarios quickly, reflecting current rate paths and liquidity stresses. Governance lag is a risk in its own right: if recalibration takes weeks, exposure keeps building in the meantime 

• Integrated decision support. When models lag, commercial teams start making decisions based on guesswork. Outputs must link to pricing, hedging, and product design. Here, IRRBB can provide a valuable lens on margin, funding cost, and balance sheet strategy.  

Dynamic modelling changes the conversation from, “What’s our exposure?” to “What’s our next move?” Regulators are already pushing for this shift, and banks must move with it, managing risk proactively rather than defensively. 

Regulatory pressure is rising

Recent stress events have hardened the supervisory tone across jurisdictions and the regulatory message is clear: static modelling won’t cut it anymore. Globally, regulators are calling for: 

• Forward‑looking, strategy‑informed modelling that links IRRBB to capital and liquidity planning 

• Dynamic modelling of non-maturity deposits (NMDs) and prepayment risk, backed by evidence‑based recalibration 

• An end to “overly static modelling”, setting expectations for real‑time responsiveness, with feedback loops between IRRBB, risk appetite, and product strategy. 

Governance frameworks are coming under similar scrutiny, with supervisors wanting to see assumptions challenged continuously. They are also testing responsiveness and decision latency, with a focus on how quickly firms update models and translate insights into action. 

Governance has become an active constraint 

Adaptability is increasingly what separates resilient banks from over-exposed ones. A less precise model that updates quickly often supports better decisions than a highly accurate model that’s slow to recalibrate. Boards and Asset and Liability Committees (ALCOs) should treat model responsiveness as a first‑order risk metric. 

Governance and culture must also evolve alongside the models themselves. Governance cycles built for a stable world now need to operate in motion, and banks’ oversight mechanisms need to keep pace. That means asking: 

• What assumptions have aged out? 

• What behavioural signals are we missing? 

• How quickly can we pivot if they change? 

The risks of getting this wrong will quickly show up in margin compression, liquidity strain, and, in extreme cases, franchise risk. 

Closing thought 

In a market where behaviour moves faster than governance, resilience hinges on how quickly banks can translate insight into action, because that’s where risk and growth now intersect. To bring that kind of agility to modelling, we believe firms must gain a deeper understanding of deposit behaviour and build an architecture that allows teams to act on those insights, rapidly and consistently.