The future of bank stress testing

December 2025

In our March 2025 article, KPMG in the UK explored how political uncertainty, geopolitical pressure and shifting global dynamics are reshaping the stress testing landscape. Since then, the pace of change has only accelerated. To understand how UK banks are responding, in Q2 2025 KPMG conducted a survey of stress testing practices, engaging large and mid-sized institutions.

A consistent theme was observed – that while regulatory stress testing frameworks are relatively mature and well-established, they are typically not agile enough to deal with emerging risks, including recent geopolitical activity. The current landscape demands agility, speed and the ability to model complex, interconnected and rapidly evolving risks. Without these enhanced capabilities, firms may make decisions based on outdated insights, miss early warning signals and face increased regulatory scrutiny.

Stress testing capabilities are at an inflexion point and now is the time to harness new approaches and technologies to support more forward-looking and integrated risk management. It’s time to put aside the familiar playbook and reshape the landscape.

What could the capabilities of the future look like?

Current capabilities will soon feel outdated, replaced by a new era of speed and precision. AI is set to transform the way stress testing operates in a fundamental way. Firms are moving beyond experimentation into practical implementation to leverage AI’s full potential, supported by robust governance and safeguards. Banks are exploring use cases including:

Seamless data cleansing and preparation;
Accelerated development and parameterisation of sophisticated scenarios;
Validation and challenge of model outputs;
Analysis of outputs against limits and thresholds with recommended management actions; and
Production of structured and unstructured reporting.

A shift is required towards a service-based operating model in which forecasting and stress testing capabilities seamlessly support functions across an organisation. Traditional barriers between risk, finance, treasury and business functions are increasingly counterproductive. The breaking down of silos can enable a more unified and optimised ecosystem of capabilities that delivers measurable time and cost efficiencies across execution, governance and reporting, for example through:

A single source of data which reduces the overall controls burden and divergence of outputs;
Fewer models and calculators to develop, validate and maintain; and
Elimination of duplicative processes with often contradicting outputs that need to be analysed, compared and explained.

The output will be a single set of results to send through governance layers, freeing up senior stakeholder time to focus on the ‘so what’.

Targeted investments can deliver upgrades to data infrastructure, replacing years of tactical fixes for data quality issues. Data remediation and acquisition will deliver more granular sector- and customer-specific data, enabling product-sensitive obligor level modelling at the required frequencies – a move away from current processes which struggle to match the pace of emerging risks. We expect to see agile and modular modelling architecture that works flexibly across different asset classes and business lines delivering product-specific insights with broad coverage of complex exposures. Automated, nimble workflows will become the norm, reducing reporting lag and freeing analytical capacity for higher-value activities.

The new capabilities outlined above can help to unlock the full value of stress testing by transforming outputs into actionable risk intelligence, embedding risk management into frontline decision-making. Today, stress testing is largely confined to limit setting and periodic risk monitoring, with outdated and infrequent data combined with process lags often rendering insights stale by the time they reach decision-makers. Looking ahead, more timely, if not real-time, risk intelligence – including analysis of third-party behaviours in stress conditions – can deliver tangible benefits, including higher earnings, through forward-looking risk management that drives better business decisions and risk-based pricing, ultimately reducing capital requirements .

The largest banks in the UK have already set aside more than £50 million (source: KPMG Stress Testing Survey October 2025) to deliver this vision over the next two years. Banks that do not make changes could soon be left behind.

Rethinking supervisory stress testing

Regulatory stress testing is also evolving to keep pace with these advancements. Regulators acknowledge that the landscape has evolved significantly since regulatory stress tests were first introduced over a decade ago. Banks today are well capitalised, and tests have consistently demonstrated that the sector is resilient against severe global stagflationary and recessionary stresses. However, bank failures serve as a stark reminder that traditional scenarios calibrated on historical events are no longer sufficient. A more dynamic regulatory approach is required to maintain industry resilience in the face of evolving risks.

In 2024, the Bank of England/PRA refreshed its stress testing framework for banks, moving to a biannual Bank Capital Stress Test (BCST), with the intervening year focused on exploratory exercises. In response to extreme and unexpected volatility in market liquidity triggered by sudden policy changes, the PRA carried out the first System Wide Exploratory Scenario (SWES) in 2023-2024. The SWES assessed interconnectedness and compounding risks across a broader range of market participants and considered whether the actions taken by individual firms would mitigate or amplify the stress. The exercise provided valuable insights to both regulators and participants.

The US Federal Reserve is reviewing its approach to the Comprehensive Capital Analysis and Review (CCAR) framework to ensure its continued credibility. In Europe, the ECB is also exploring alternative approaches and planning a reverse stress test focused on geopolitical risks. Unlike traditional stress tests, this exercise would provide participants with the impact of stress and asking them to identify and explain the events that could take them to such outcomes.

It’s little surprise that, in the UK BCST, banks’ capital positions have once again proved robust, supporting the reduction of the system-wide Tier 1 capital benchmark rate to 13%. Looking ahead to 2026, the focus will be on understanding the broader risks and dynamics of private markets, with the next SWES exercise set to test the resilience of the ecosystem through a macroeconomic scenario.

How KPMG in the UK can help

KPMG in the UK is dedicated to supporting clients through insights and leading practices that can help drive success. Our knowledge of industry trends and regulatory changes can help firms navigate complexity and capitalise on new opportunities. We have supported banks since the first UK concurrent stress test in 2014 and have access to extensive international experience from across the global network of member firms. To discuss any of the themes above in more detail, please reach out.