Comparison of the Bank of England’s 2025 Stress Testing Framework vs 2024 and 2022 Frameworks

Introduction and Background

On the 24th March 2025, the Bank of England (BoE) published the key elements of the 2025 Bank Capital stress test. The BOE has progressively refined its bank stress testing framework over the past decade. In 2022, it formalized a Stress Test Data Framework (STDF) to standardize data submissions for its Annual Cyclical Scenario (ACS) stress tests. By 2024, reflecting increased bank capital levels and lessons learned, the BoE announced an updated approach to stress testing effective from 2025 onward. A cornerstone of this new approach is the replacement of the annual ACS exercise with a Bank Capital Stress Test (BCST) conducted every two years. The BCST serves as the successor to the ACS, marking a shift in frequency and design of cyclical stress tests. At the same time, the BoE continues to run exploratory scenario exercises to probe structural or emerging risks, building on the Biennial Exploratory Scenarios (BES) used in earlier frameworks.

This report provides a detailed comparison of the BoE’s 2025 stress testing framework with the prior frameworks – in particular, the 2024 approach (a transition year) and the original 2022 STDF/ACS framework. It outlines all key changes introduced in 2025 relative to 2024, explaining the context and rationale. The comparison covers changes in test frequency and scheduling, scenario design, methodology, bank participation, capital adequacy evaluation, and public disclosures. A side-by-side summary table is included below for quick reference, followed by in-depth sections on each aspect. All information is drawn from official BoE publications and guidance to ensure accuracy and a factual focus (avoiding speculative commentary on implications).

Summary of Key Changes in 2025 vs 2024 and 2022 Frameworks

The table below highlights the key differences between the 2025 stress test framework and the previous frameworks (2024 and 2022) across major dimensions:

Aspect	2022/23 Framework (ACS & STDF)	2024 Framework (Transitional)	2025 Framework (BCST)
Test Frequency & Cycle	Annual ACS + biennial BES exploratory scenarios	No ACS; desk-based stress test + SWES (System-Wide Exploratory Scenario)	Biennial BCST with full bank participation; lighter desk-based/ICAAP tests in alternate years; ad-hoc exploratory exercises continue
Scenario Design	Severe macro & market shock ACS (5-year horizon) + separate BES scenarios (e.g. climate)	Two adverse desk-based scenarios; SWES explored structural risks	One coherent “tail risk” scenario (e.g. supply shock); 5-year horizon; adjusted for IFRS 9 earlier provisioning; exploratory scenarios remain separate
Methodology & Execution	Bank-led using internal models and STDF templates; BoE oversight; IFRS 9 transitional relief applied	BoE-led internal exercise; no bank submissions; based on supervisory models; banks used ICAAP with similar scenarios	Hybrid: Bank submissions + BoE top-down analysis; uses 2022 STDF templates; IFRS 9 fully loaded (no transitional relief)
Bank Participation	8 major/systemic banks (incl. Virgin Money UK); smaller banks monitored via other tools	Same major banks implicitly considered; no active participation; broader inclusion via SWES	7 largest/systemic banks (excl. Virgin Money UK which was acquired by Nationwide); smaller firms assessed separately
Capital Adequacy Evaluation	Hurdle rates used; outcomes informed PRA buffers; IFRS 9 transitional relief factored in	No pass/fail; results informed CCyB setting; assessment aligned with mostly IFRS 9 fully loaded basis	No mechanical pass/fail; PRA buffers set based on results; CET1 drawdowns allowed to 2% CCyB; IFRS 9 fully integrated
Disclosures & Transparency	Full disclosure of system and bank-specific results; detailed scenario assumptions and capital paths published	Only aggregate results disclosed in Financial Stability Report; no firm-specific outcomes	Restores full disclosure: includes each bank’s CET1 low point vs. requirement; clearer interpretation of buffer usage under the new approach

Table: Side-by-side comparison of the Bank of England’s 2022, 2024, and 2025 stress testing frameworks across key dimensions. ACS = Annual Cyclical Scenario; STDF = Stress Test Data Framework; BCST = Bank Capital Stress Test; BES = Biennial Exploratory Scenario; CCyB = Countercyclical Capital Buffer; FPC = Financial Policy Committee; PRC = Prudential Regulation Committee; PRA = Prudential Regulation Authority.

Test Frequency & Cycle: Transition from Annual ACS to Biennial BCST

One of the most significant changes in 2025 is the move from an annual to a biennial stress test cycle, effectively replacing the ACS with the BCST. Under the previous approach (established in 2015), the BoE ran an ACS every year to gauge banks’ resilience to financial-cycle risks, and roughly every two years ran an exploratory scenario to probe other risks. The ACS was central to setting capital buffers annually. By contrast, starting 2025, the BoE will conduct the Bank Capital Stress Test every other year, with no annual ACS in between. This BCST is explicitly described as “the successor to the Annual Cyclical Scenario exercises”. In practice, this means 2025 hosts a full stress test with bank participation, 2026 will not have a bank-run capital test (focusing on lighter analyses instead), 2027 will have the next BCST, and so on.

Context for the Change: This shift was motivated by a recognition that the banking system is much better capitalized now than a decade ago, so running the identical heavy exercise every year yielded diminishing returns. Banks themselves had given feedback that an annual cycle was burdensome and might crowd out other risk management efforts. The BoE noted that the marginal benefit of yearly full-scope tests had diminished as capital levels rose and capabilities improved. By moving to a biennial cycle, the BoE aims to reduce the burden on participating banks and free up bandwidth to examine a wider range of risks over time. Indeed, the updated approach “represents a material efficiency gain… and supports the UK banking sector’s competitiveness and growth” while keeping resilience standards unchanged.

Intervening Year Exercises: In years without a BCST, the BoE will still conduct stress test exercises, but these will be “less burdensome” desk-based tests or targeted analyses rather than full concurrent exercises. For example, 2024 served as a prototype of this approach: the BoE ran an internal desk-based stress test with two scenarios and published aggregate results. In alternate years going forward, similar exercises could include BoE staff modelling a scenario using banks’ data (without requiring full submissions) or asking firms to run specific targeted mini-stress tests on particular portfolios. The BoE has also indicated it may leverage banks’ own ICAAP stress test results during off-years. These off-year tests supplement the biennial BCST to ensure continuous monitoring of cyclical risks, albeit through more flexible means.

Continued Exploratory Scenarios: Separately, the BoE will continue to conduct exploratory scenario exercises for risks not tied to the credit cycle (for instance, climate change, fintech disruptions, or market liquidity freezes). Under the old framework, such Biennial Exploratory Scenarios (BES) ran alongside the annual ACS (e.g. the 2021 Climate BES). Under the new approach, these are not strictly on a fixed schedule, but the Bank “will take into account the risk environment and the sequencing of other tests” to decide when to run them. The 2024 System-Wide Exploratory Scenario (SWES) is an example – it was the BoE’s first exploratory test involving multiple parts of the financial system, conducted in late 2024. Going forward, we might see exploratory scenarios in years where there is no BCST (to even out workload), but the timing remains at the BoE’s discretion. Crucially, these exploratory exercises can involve a broader set of participants beyond the core seven banks, depending on the risk being examined. For instance, a climate scenario might involve insurers and overseas banks, whereas a fintech disruption scenario could involve payment systems, etc. The results of exploratory tests are generally published only at aggregate level (focused on system findings), as was done for the climate BES in 2021.

In summary, 2025 marks a shift to a two-year cycle for major stress tests. The ACS – which was the annual concurrent stress scenario – has been replaced by the BCST as a biennial exercise. This change is designed to maintain the robustness of stress testing while improving efficiency and allowing the BoE to investigate a wider array of risks over time. The BoE has stressed that this will not weaken standards: the BCST is still a rigorous “tail risk” test and, taken together with interim analyses and exploratory scenarios, will continue to fully support the FPC’s and PRA’s objectives for financial stability.

Scenario Design: ACS vs. BCST and Related Changes

Under the ACS framework (2022 and prior), the BoE’s scenario design followed a consistent philosophy: craft a severe but plausible adverse scenario linked to the financial cycle, assuming banks keep lending throughout the stress. The ACS scenarios typically involved a UK and global recession, rising unemployment, asset price crashes, and sometimes specific stresses like misconduct costs or trading book losses. For example, the 2022/23 ACS (run in 2022) was calibrated on the assumption that banks “satisfy the demand for credit from UK households and businesses throughout the stress” – i.e. they do not deleverage or cut off lending. This assumption forces the scenario to be tougher: banks can’t avoid losses by shrinking, so the macro shock must be sized such that even with continued lending, the system is pushed to its resilience limit. The Key Elements of the 2022 ACS laid out a scenario with a five-year horizon starting in 2022 Q3, featuring a sharp rise in unemployment, a GDP contraction, falls in property prices, etc., alongside a “separate stress of misconduct costs.”. Indeed, since the mid-2010s, BoE stress scenarios have often included an additional overlay for operational-risk losses (misconduct fines or redress costs) that hit banks during the scenario. The 2022 ACS explicitly had a Section 3.5 Misconduct stress where banks had to project costs for known issues beyond existing provisions. Similarly, market risk was stressed via a set of traded risk shocks (e.g. a severe move in interest rates, credit spreads, FX, etc.), for which the BoE provided a “Traded risk scenario” dataset to participants. In summary, the ACS scenario design combined a macro stress (credit risk focus), market shock (trading book/valuation focus), and misconduct/operational risk stress – all unfolding over a multi-year horizon.

The 2025 BCST scenario inherits much of this design philosophy, but with some updates and refinements:

Macro Scenario: The 2025 Bank Capital Stress Test scenario is likewise a tail-risk macroeconomic scenario, described as deep simultaneous recessions in the UK and global economy, with large falls in asset prices and higher global interest rates. In other words, it’s a broad-based severe downturn scenario not unlike prior ACS exercises. The specific narrative in 2025 is an aggregate supply shock leading to stagflationary conditions (output down, inflation and rates up). The Key Elements of 2025 document details the exact paths: for instance, UK GDP falls ~5%, world GDP falls ~2%, UK unemployment rises to ~8.5%, etc., at the peak of stress. The horizon remains around five years (the scenario runs into 2029 for the 2025 test). Crucially, the scenario continues to be “calibrated on the assumption that banks continue to support the economy, including through lending in stress”. This is explicitly stated in the participant guidance, reinforcing the same assumption as the ACS: banks must meet credit demand at rates consistent with the scenario, rather than avoid losses by cutting exposure. So, the fundamental design – a countercyclical severe downturn with no bank-driven credit contraction – is unchanged.
Timing of Stress Peaks: One deliberate change in 2025 is the timing of the scenario’s worst points. The BoE “reviewed the timing and transmission of shocks” and delayed the peaks/troughs of key variables (unemployment, GDP, house prices, etc.) compared to previous stress tests. In practice, this means the recession unfolds slightly later and more gradually in the scenario timeline, causing some credit losses to be recognized later rather than immediately. The reason is to align with IFRS 9 dynamics: since banks now must take provisions earlier (because under IFRS 9 they book expected losses as soon as a downturn is forecasted), if the scenario’s severe phase is slightly delayed, the loss recognition is spread out, avoiding unrealistically front-loaded hits. The BoE notes this change should not materially reduce severity, but it changes the shape of loss accumulation.
Severity Calibration: The size of shocks in the scenario is calibrated to maintain comparable overall severity to past exercises (given the same risk tolerance). However, the BoE simplified its method by incorporating more historical data on downturns. This led to some start-to-trough changes being “somewhat smaller” in the 2025 scenario than they would have been under the old method. For example, if previously the model might have shocked a variable by, say, –30% to reach an unprecedented level, the new method might use a slightly smaller shock if historical analysis suggests the old approach overshot. The BoE emphasizes that any reduction in individual variable movements is minor and does not significantly affect the aggregate capital drawdown in the stress. Essentially, they fine-tuned the calibration to ensure consistency with actual observed correlations and extremes, simplifying where possible. This makes the scenario design process more straightforward and data-driven, while still ensuring the test is very severe (worst-case but plausible).
Credit Portfolio Specifics: Another tweak in scenario design is in consumer credit loss assumptions. In previous ACS exercises, the BoE had taken a very conservative view on UK consumer credit (e.g. credit card and unsecured loan) losses, partly because it believed low default rates in benign years were mainly due to a good economy, not better underwriting. By 2025, after reviewing data, the Bank now credits some improvement to underlying quality. Thus, projected loss rates on consumer credit in the 2025 stress are allowed to be a bit lower than in past tests (all else equal). This doesn’t change the overall macro scenario, but it means when banks model their loan losses, the impairment on credit cards, personal loans, etc., may be slightly less harsh relative to the macro variables than previously assumed. It’s a recognition that, for example, credit scores, lending standards, or portfolio shifts have structurally reduced unsecured lending risk to some extent.
Misconduct and Traded Risk Elements: The 2025 test explicitly retains the three-part structure of (1) macroeconomic scenario, (2) traded risk scenario, and (3) misconduct stress. The news release announcing the test highlighted that it “has three elements” including those components. Thus, similar to 2022, banks in 2025 must estimate potential misconduct costs in addition to the macro impacts. The guidance likely asks banks to consider any known conduct issues and provide a stressed cost estimate (as was done in 2022). Likewise, a traded risk scenario (usually a set of instantaneous shocks to market prices and rates) is provided – indeed the BoE published a spreadsheet of “Traded risk scenario for the 2025 test” alongside the macro variable paths. This ensures that banks with trading operations apply a consistent market shock (e.g. equity prices drop X%, credit spreads widen Y bps, etc.) during the stress. Notably, these market shocks often occur early in the scenario (year 1), reflecting a market crisis aspect within the broader macro downturn.
Exploratory Scenario Design: Although not part of the BCST per se, it’s worth noting how exploratory scenarios are handled in design terms post-2025. The BoE will design those tests to suit the risk being examined – for example, the 2021 Climate BES had a 30-year horizon to assess long-term climate risks, whereas a future fintech disruption scenario might be shorter term but involve different participants. The updated framework doesn’t lock in a particular design for these; it simply preserves flexibility, with the understanding that such scenarios are not tied to the credit cycle and may require bespoke metrics. When they occur, they will be designed under guidance of FPC/PRC like other tests and coordinated with the timing of BCSTs so as not to overload firms at the same time.

In summary, the scenario design in 2025’s framework remains focused on severe macro-financial stress with ancillary risk components (market and misconduct), much like the 2022 ACS, but adjustments have been made to reflect the new accounting regime and insights:

The Annual Cyclical Scenario (ACS) used up to 2022 has been superseded by the Bank Capital Stress Test (BCST) scenario in 2025, which serves the same purpose but runs biennially.
The BCST scenario’s calibration is slightly refined (delayed peaks, updated shock sizing, revised loss assumptions) to balance IFRS 9’s effect while keeping the overall stress severity in line with the BoE’s unchanged risk appetite for bank resilience.
Key scenario features like countercyclical severity, tail-risk scope, inclusion of traded risk and misconduct costs, and five-year horizon continue from prior frameworks.
The scenario design still explicitly assumes banks do not pull back lending, which is central to both the old and new frameworks’ philosophy (ensuring the test probes banks’ ability to support the economy through a shock).

Methodology and Data Framework Changes

The methodological approach of the BoE’s stress testing encompasses how the test is executed – i.e., who does the modelling, how data is gathered, and how certain rules (like accounting) are applied. There have been some important methodological and data-related changes from 2022 to 2025:

Use of Banks’ Models vs BoE Models: Under the concurrent stress testing paradigm since 2014, the BoE has generally followed a “distributed modelling” approach – participating banks run their internal risk models to project losses, revenues, and capital under the common scenario, and then the BoE reviews and challenges these projections using its own analytical tools. This was true in 2022: each of the eight banks used its own models for credit risk (PD/LGD projections), market risk, net interest income, etc., and filled out the prescribed STDF templates with results. The BoE would then aggregate the data and apply overlays or adjustments as needed before finalizing results. In 2025, with the reinstated BCST, this approach continues – banks are again primary contributors of projections. The 2025 Guidance for Participants underscores that banks must submit detailed results (capital positions, P&L, etc.) and follow the STDF Manual instructions when doing so. The BoE staff then use these submissions, combined with BoE models, to assess the outcomes. This collaboration between firm-specific granular modelling and BoE top-down analysis remains a cornerstone of the methodology.

By contrast, the 2024 desk-based test deviated from this: it was run almost entirely by BoE staff without asking banks to produce stress forecasts. The BoE essentially leveraged supervisory data and perhaps the banks’ earlier ICAAP projections to simulate the impact of two scenarios on bank capital. This is why no STDF data collection occurred in 2024 for ACS – there was no ACS. The BoE noted such desk-based exercises “rely primarily on the Bank’s own estimates of the impact of stress scenarios”. This method is inherently less precise for individual banks (since it may use standardized or less granular models), but it was deemed sufficient for a one-year gap. The methodology in 2024 was also experimental in nature, testing how to incorporate multiple scenarios and use them for CCyB decisions without doing a full concurrent test.

Stress Test Data Framework (STDF): The STDF was introduced around 2021-2022 as a standardized approach to data collection for stress tests. The 2022 STDF Manual provided an overview and instructions for all the Excel templates banks must submit. These templates cover everything from detailed loan loss projections, risk-weighted asset (RWA) calculations, balance sheet and income statement forecasts, to capital position reconciliation, among others. The STDF also includes a dictionary defining each data field and validation checks to ensure internal consistency. In essence, the STDF is the infrastructure enabling concurrent stress testing data to be collected consistently across banks.

In 2022, the STDF was new, so the BoE had just completed a “comprehensive review” of data needs for the ACS from 2022 onwards. The Manual details submission frequency, file formats, quality checks, etc., and even outlines that submissions should include both baseline and stress scenario projections for comparison. One notable point: even though IFRS 9 transitional arrangements were still in effect then, the STDF instructions told banks to report certain capital metrics on an “IFRS 9 fully loaded basis” (i.e. as if no transitional relief) to give the BoE full insight. Meanwhile, other parts of the data could be on a transitional basis – this nuance ensured the BoE could see the difference IFRS 9 rules made. The STDF in 2022 and 2023 was heavily focused on the ACS exercise (since that was the main stress test happening).

For 2025, the BoE decided not to overhaul the STDF templates for the new framework. Instead, it kept the 2022 templates, dictionary, and manual unchanged (except for a minor update to one “Basis of Preparation” document). The rationale was to minimize costs and effort for banks – if forms and definitions changed, banks would have to adapt their internal systems to populate them, which can be labour-intensive. So, somewhat interestingly, the 2025 stress test still uses fields labelled “Annual Cyclical Scenario” or “ACS” in the templates, even though the exercise is now called BCST. The BoE explicitly noted that it “has not changed any references to the ACS in the templates, STDF dictionary or STDF manual” for 2025. Banks are simply instructed to treat “ACS” as referring to the 2025 Bank Capital Stress Test scenario for reporting purposes. This continuity indicates that the data points being collected (loan losses, incomes, capital ratios, etc.) are essentially the same as before. There were no new templates introduced specifically for exploratory scenarios or the like – presumably, if an exploratory test requires data, the BoE will create a bespoke request at that time.

IFRS 9 Accounting Treatment: Methodologically, one of the biggest changes from 2022 to 2025 is the accounting standard environment. IFRS 9 (effective 2018) changed how banks provision for credit losses, requiring an expected-loss approach. To avoid a sudden drop in regulatory capital when IFRS 9 was implemented, regulators allowed a phased transitional arrangement (often 5 years) where banks could “add back” a portion of increased provisions to their capital calculations, tapering that relief over time. This transition period ended by January 2023 (in line with the schedule, relief reduced to zero by end-2024). Therefore, the 2025 stress test is the first concurrent test with IFRS 9 fully in force and no transitional cushion.

This has methodological implications:

In earlier stress tests (2018-2022), when banks projected credit losses, IFRS 9 would make them recognize losses faster (earlier in the scenario) than the old standard (IAS 39) would. But because of transitional rules, some of that impact didn’t hit their regulatory capital ratios fully – the rules allowed adding back a decreasing percentage of “stage 1 and 2” provision increases. The BoE in stress tests would consider those rules when judging results.
In 2025, banks’ stressed capital trajectories will reflect full IFRS 9 impact – meaning potentially larger, front-loaded hits to capital ratios as soon as the scenario deteriorates. The BoE adjusted its calibration (timing of peaks, as mentioned) to avoid double counting this effect. Moreover, from a methodology perspective, the BoE has stated it “intends to make changes to the stress-testing framework that ensure the resilience that comes with earlier provisioning is recognised”. The idea is that IFRS 9, by forcing banks to provision early, actually gives a benefit to resilience (banks start the real stress with more provisions and thus are better covered against losses). So, the BoE doesn’t want to penalize banks for that in the test; it wants to acknowledge that earlier loss absorption improves outcomes. Concretely, one methodological adjustment was assuming a higher CCyB (2%) at the start of the test, as discussed in the capital evaluation context, to let capital ratios fall further without “failing”. Another is that the BoE will consider the level and nature of provisions under IFRS 9 when using the results – essentially a reminder that it will look qualitatively at whether a bank’s big capital drop is partly because it was conservative in recognizing losses upfront (which is actually a good thing in reality).

In summary, the methodology in 2025 is adapted to fully incorporate IFRS 9:

There is no transitional crutch, so the stress scenario design and result interpretation were tweaked to account for the “earlier provisioning” phenomenon.
The BoE will likely perform analyses (as hinted in Box A: Incorporating IFRS 9) to see how much more provisioned losses are hitting capital relative to past tests and then use judgment for buffer calibration.

Quality Assurance and Process: The STDF manual details the submission process and data quality checks. In 2022, banks had to follow a strict timeline to submit initial results, answer queries, possibly resubmit adjusted figures after BoE feedback, etc. That timeline usually spanned ~3-4 months from scenario launch to final submission. The 2024 approach essentially skipped this cycle. For 2025, we expect a return of the full process: BoE released the scenario on 24 March 2025; banks will deliver initial results (likely mid-year), BoE will engage in a “challenge process” to discuss outliers or inconsistencies, and final results will be compiled in the fall. The submission frequency under STDF is typically a single comprehensive submission for the scenario, plus possibly some interim data. The 2022 manual section C.1 (Submission frequency) likely indicates one primary submission and provisions for resubmissions if needed. The 2025 guidance confirms “Submission instructions are outlined in the STDF manual” and banks must follow those for structured (templates) and unstructured (qualitative answers) data. It also explicitly says “No changes have been made to core templates” in 2025, highlighting that the process and format remain as in 2022.

In conclusion, methodologically the 2025 framework:

Returns to bank-run models + STDF submissions, after a one-year deviation in 2024 where it was BoE-run.
Continues to use the STDF 2022 infrastructure, affirming continuity in data requirements.
Integrates IFRS 9 fully, with adjustments to scenario timing and supervisory evaluation to keep the test fair and consistent with prior severity.
Maintains the principle that results are an input to judgment, not an automatic determinant – as stressed in methodology notes that there is no mechanical link to setting capital, and many inputs are considered.

Meanwhile, the 2024 exercise demonstrated a new capability: BoE can, when needed, run internal stress analyses without the full STDF process. This is now formalized as part of the framework for off-years, giving flexibility in methodology (e.g., using “multiple scenarios, different levels of granularity” or targeted tests on subsets of balance sheets). This agility is a methodological broadening: rather than one fixed way to do stress tests, the BoE can choose the tool appropriate for the year and risk in question, whether that’s a bank-led detailed test (like 2025) or a simpler desk study (like 2024).

Bank Participation and Scope of Coverage

Participating Institutions in 2022 (ACS): The original ACS framework included the largest UK banking groups and occasionally large building societies. In the 2022/23 ACS, eight banks/building societies participated. These were specifically: Barclays, HSBC, Lloyds Banking Group, Nationwide Building Society, NatWest Group, Santander UK, Standard Chartered, and Virgin Money UK. This list covers all the major UK domestic systemically important banks (DSIBs) and one mid-sized challenger bank (Virgin Money) which was included likely due to its significant retail lending market share after merging with CYBG. The inclusion of Virgin Money in 2022 marked an expansion of scope compared to earlier years (for instance, earlier in the 2010s, the test had seven firms, Virgin Money was added later). The BoE’s approach for deciding participants was (and remains) based on systemic footprint – e.g., share of lending to UK households and businesses, overall balance sheet size, and importance to financial system. In 2022, the coverage of those eight firms captured an estimated 80%+ of UK bank lending, ensuring the system’s core was tested. The BoE also sometimes assessed ring-fenced subgroups within those banks: for example, HSBC UK (ring-fenced bank) vs HSBC Group, etc., if needed for granular supervision. However, external disclosures and the capital setting were usually at the consolidated group level.

2024 Transition – No Bank Submissions: For the reasons discussed, 2024 did not have an official list of “participating banks” in the same sense, since no bank had to submit results for an ACS. However, the desk-based test was explicitly said to be “of the UK banking system… as represented by the major UK banks and building societies”. This implies that the BoE’s analysis implicitly covered the same institutions (the ones that would normally be tested). The BoE likely took the balance sheets of those major eight firms and applied the scenario impacts to them through its models. Therefore, while not an ACS, the scope in 2024 still encompassed the big eight. In addition, the BoE took the extra step of providing the adverse scenarios publicly so that smaller banks and building societies outside the concurrent test could use them for their own internal stress testing. On 27 June 2024, the PRA published two scenarios for this purpose, which were identical to the BoE’s desk-based test scenarios. This was a way to extend the reach of the exercise to non-participating firms in a consistent manner. Also in 2024, the System-wide exploratory scenario (SWES) included not just banks but other financial institutions, although details of which entities participated were not fully public (it likely involved key banks, insurers, perhaps central counterparties, etc., to study system interactions).

Participants in 2025 (BCST): The BoE has clarified that the 2025 Bank Capital Stress Test is being run for “the seven largest and most systemic UK banks and building societies.” While the official communication did not list the names, it is almost certain this refers to the same set as 2022 minus Virgin Money UK (the smallest of the prior eight). Thus, the likely participants are Barclays, HSBC (Holdings plc’s UK consolidation, including HSBC Bank plc and HSBC UK bank), Lloyds Banking Group, Nationwide Building Society, NatWest Group, Santander UK (UK part of the Santander group), and Standard Chartered. This covers essentially all institutions designated as major domestic players or global systemically important (HSBC and Standard Chartered are GSIBs with UK presence). The BoE’s updated approach statement notes that participation may change over time based on share of lending and systemic importance, and that any new bank added would get 12 months’ notice. That implies the list isn’t permanently fixed; for instance, if another mid-tier bank grows or if a new large player enters the market, they could be included in future tests (with advance warning).

Non-Systemic Firms: The new framework explicitly addresses how smaller banks and foreign bank subsidiaries are handled. It states that insights into non-systemic banks’ resilience will be delivered to the FPC/PRC by using existing tools and possibly the banks’ own stress testing (ICAAPs). In practice, this means the PRA will continue to run its annual ICAAP supervisory stress for each bank (where each bank must demonstrate its capital adequacy in scenarios as part of Pillar 2). The BoE can aggregate or review those to ensure smaller institutions are sound, even if they’re not in the BCST. Additionally, for UK subsidiaries or branches of foreign banks, the BoE/PRA will rely on home regulators’ stress tests and ongoing supervision. For example, a big US or EU investment bank’s London branch won’t be in the BCST, but the BoE coordinates with the Fed or ECB respectively to understand how that institution fares under stress. This approach wasn’t new in 2025, but the updated framework reiterates it as part of the holistic coverage of the banking sector.

Exploratory Exercise Participation: When it comes to exploratory scenarios (like climate), participation is determined by relevance and proportionality. The BoE indicates it will decide which banks (or even other financial firms) should take part based on whose business models are most exposed to the scenario and whether it’s appropriate (cost/benefit) to involve them. For instance, the 2021 Climate BES included the same seven major banks and three large insurers, since climate risk was salient to both sectors. A future exploratory scenario on, say, digital finance might involve a different mix of firms. The framework also commits to giving appropriate notice to firms invited for these exercises and ensuring the scope of data requested is proportionate. This addresses a criticism that some exploratory scenarios in the past (like the 2017 BES on consumer credit, if any, or 2021 climate) were as data heavy as the ACS, adding burden.

Summary: The scope of the stress testing framework in 2025 remains the large, systemically important banks, with a conscious narrowing to exclude the smallest of the previously tested banks. This aligns with the goal of focusing resources on the most critical institutions for UK financial stability, while using other tools for the rest. To put it succinctly:

In 2022, 8 banks/BS participated in ACS.
In 2025, 7 banks/BS participate in BCST (a reduction by one, due to Virgin Money being acquired by Nationwide).
The BoE’s approach going forward is to adjust the participant set only if warranted by changes in the banking landscape, and with ample notice.
Coverage of lending remains high – those 7 banks cover the bulk of UK loans. The BoE has stated the coverage of the banking sector’s lending is a factor in inclusion (ensuring the test captures most of the system’s credit exposure).
All told, this ensures that the biennial stress test keeps its focus on systemic risk core, while other institutions are not ignored but are stress-tested in proportionate ways outside the concurrent framework.

Capital Adequacy Evaluation and Outcomes Assessment

The ultimate purpose of these stress tests is to evaluate whether banks hold enough capital to withstand severe shocks while continuing to perform their functions. Over time, the BoE has evolved how it uses stress test results in its capital framework – moving from a pass/fail mindset to a more nuanced capital planning tool. The 2025 framework further refines this evaluation process, especially in light of IFRS 9.

2014–2017: Pass/Fail with Hurdle Rates. In the early years of BoE stress testing (after it became regular in 2014), the results were directly tied to hurdle rates. The BoE would set a CET1 capital ratio threshold for each bank, typically comprising the minimum CET1 requirement (Pillar 1) plus the bank’s Systemic Risk Buffer or GSIB buffer, and sometimes plus Pillar 2A requirements (depending on the year’s framework). Banks also had to meet a leverage ratio threshold. If a bank’s projected ratio in the stress fell below its hurdle, it could be required to take remedial action (like raising capital or reducing risk exposures). This was akin to a test “failure.” For example, in the 2016 stress test, two banks (RBS and Standard Chartered) fell below one of their hurdle rates and were required to improve capital plans.

2018–2022: Integrating Stress into Buffer Setting. After 2017, the BoE, in line with changes to UK capital framework, removed the explicit pass/fail hurdle in favour of using results to size each bank’s PRA buffer (Pillar 2B). The PRA buffer is an extra layer of capital the supervisor requires above minimums and conservation buffer, intended to cover risks either from stress test outcomes or poor risk management. The BoE made clear that stress test results are a key input, but “not a mechanical link” to capital requirements. Essentially, the worst loss a bank suffers in the stress (beyond its capital conservation buffer) informs how big its PRA buffer should be – the idea being that if a bank barely stays above minimum in the stress, perhaps it should hold more going forward. However, other factors (like qualitative risk management, scenario realism, etc.) are considered too. By 2022, it was established that if a bank dipped into its Combined Buffer Requirement (CBR) in the stress (i.e., went below the sum of minimum plus capital conservation buffer plus any systemic buffer), that would not automatically trigger a fail; instead, it would lead to supervisory scrutiny of whether its capital planning buffer is adequate.

The BoE’s 2022 ACS results communication highlighted that transitional IFRS 9 impacts were considered in evaluating capital adequacy. Specifically, since banks had that temporary capital relief, the BoE effectively looked at two sets of numbers: with and without transitional add-backs. This was to ensure that the PRA buffer setting took into account the fully loaded view (because by 2025, those transitional supports would be gone). The Key Elements of 2022 also indicated the test was judged against an “unchanged risk tolerance” for the system – meaning the severity was such that if banks met the outcomes, it satisfied the FPC that banks were resilient enough.

Countercyclical Capital Buffer (CCyB): Another macroprudential angle is the use of system-wide results to inform the CCyB. The CCyB is a capital buffer that the FPC can raise in good times (to make banks build extra capital) and cut in bad times. The FPC has said its “standard” CCyB for a neutral environment is 2% (of risk-weighted assets). In practice, the FPC was raising the CCyB to 2% during 2022-2023 as the economy recovered from COVID. Stress test results help the FPC decide if the banking system can handle a higher CCyB. If the stress test shows banks still have ample capital headroom, the FPC is comfortable setting the CCyB at the standard rate (or higher if needed). Conversely, if the stress revealed vulnerabilities, the FPC might hold CCyB at a lower level so as not to overburden banks. In late 2024, the FPC did have a 2% CCyB in place (effective July 2023) and was considering future moves. The BoE’s updated approach explicitly mentions that the BCST will be used to inform setting of capital buffers for the system and individual banks, which includes the CCyB (system buffer) and PRA buffers (bank-specific).

2025 Evaluation: Accepting Deeper Drawdowns (IFRS 9 effect). With IFRS 9 fully in play, expected losses are recognized sooner in a downturn. This means in a stress test, banks might show a steep capital ratio drop early on, then perhaps recover towards the end if provisions were conservative. The BoE’s stance is that this earlier hit actually makes banks safer in reality (because they provision upfront and can then weather the storm with those provisions). To reflect that, the BoE is basically saying: we will tolerate a larger capital ratio drawdown in the scenario than before, without concluding the bank is undercapitalized, as long as it’s consistent with the same underlying risk tolerance.

One concrete measure is the assumption of a 2% CCyB in the scenario’s starting point. In a real stress, the FPC would cut the CCyB to 0% to free up that 2% for absorption. So, if banks go into the test with an extra 2% buffer, it’s understood they can use it. Publishing each bank’s “low point” vs its minimum requirement directly addresses this: for instance, if a bank’s CET1 minimum (plus Pillar 2A) is 8% and it has a 2% CCyB and 1% systemic buffer, its total requirement in normal times is 11%. In the stress, if its CET1 falls to 7%, that’s 4 percentage points below the usual requirement, but perhaps only 1 point below the 8% hard minimum. The BoE is signalling that falling below the combined buffer (11%) down to maybe, say, 7% is not automatically a concern if it reflects the intended use of buffers. Instead, they will look at why it fell and whether the bank still meets the absolute minimums. By publishing 7% vs 8% minimum, they make clear the bank stayed above its 8% minimum by -1% (actually dipped slightly below) in that hypothetical, but that could be deemed acceptable given the circumstances. Each bank’s situation will feed into how the PRA sets its PRA buffer. If a bank went significantly below minimum (which is not expected unless something went very wrong or the bank started with barely above minimum capital), that would be a red flag.

The communication around 2025 emphasizes that the changes are “consistent with an unchanged FPC and PRC risk tolerance”. What is this risk tolerance? Essentially, it’s the degree of economic severity the committees want banks to be able to withstand without failing. If previously it was, say, a 5% GDP drop and X% unemployment spike leading to y% system capital drawdown, that hasn’t changed. What changed is IFRS 9 might cause that y% drawdown to show up differently in timing. So, the BoE adjusted technical factors (timing of scenario, etc.) to keep the effective “tightness” the same. The unchanged tolerance phrase also serves to assure that moving to a 2-year cycle and giving more room for drawdown isn’t a loosening of standards – it’s just recalibration.

PRA Buffer Setting: The PRA buffer (formerly Pillar 2B) for each bank will continue to be set after each stress test. The PRA looks at the difference between the bank’s starting capital and its stressed low point (taking into account that regulatory minimums were meant to be met). Typically, if a bank’s capital in stress falls to, say, 200 bps above its minimum requirement, the PRA might set a PRA buffer roughly equal to that 200 bps (so that going forward the bank holds that extra margin). If IFRS 9 causes a 100 bps bigger drop than before, the PRA might initially say “okay, we see a bigger drop, but we expected that – maybe we don’t increase the buffer fully by that amount because part of that drop is ‘usable’ buffer (like CCyB release).” It’s somewhat complex, but the key point is the PRA will use judgment. The BoE statement confirms there is no formula – “the stress test is one of a range of inputs and there is not a mechanical link” and “supervisory judgement includes an assessment of any other relevant information”.

However, historically, the PRA buffer was basically equal to the “impact of the stress scenario, over and above the capital conservation buffer and CCyB”. The text from the Key Elements 2025 says: In setting the PRA buffer, one input is the excess amount of capital that would be needed over and above the CCoB and relevant CCyB to withstand a severe stress. So practically, they look at how far below the combined buffer (excl. CCyB if CCyB is meant to be usable) a bank went. For example, if a bank started with 14% CET1, had a minimum of 8%, CCoB 2.5%, CCyB 2%, total 12.5% requirement plus PRA buffer maybe 1.5% = 14% total requirement. If in the stress it goes to 7%, then it used 5.5% of buffers. That overshot the CCoB + CCyB (4.5%) by 1.0%. That 1.0% might inform increasing its PRA buffer by ~1% (or if it already had one, adjusting to that level). The BoE also factors in qualitative aspects (if poor risk management contributed to losses, they might add an extra cushion).

Disclosure of Low Point vs Requirement: In prior disclosures, BoE usually said “no bank fell below its hurdle rate” or if one did, they’d comment on actions. For 2025, they plan to publish each bank’s low point and the sum of its minimum requirement and systemic buffers. For example, they might publish a table: Bank A: minimum requirement 7.5%, low point 8.0%; Bank B: min requirement 9.5%, low point 9.0%, etc. This way, if Bank B’s low point is 0.5 below its requirement, it’s transparent. The BoE indicated this is because “it is possible a bank’s capital position in stress could fall below… its systemic buffer without the PRA judging it needs more capital”. So, they want the public and analysts to see that explicitly and understand the context, rather than misinterpret it as a failure.

In summary, the capital adequacy evaluation in 2025 is characterized by:

No binary pass/fail trigger; instead, a continuum of results that inform capital buffers (as was the trend since 2018).
A recognition that using buffers is acceptable in a stress – the test is designed to use them. The FPC expects banks to “absorb rather than amplify shocks” by drawing on capital if needed.
A commitment to maintain overall stringency (unchanged risk tolerance), meaning if banks could pass before, they should similarly pass now, albeit with different optics due to accounting.
Greater transparency on individual outcomes to bolster credibility: publishing low points vs requirements allows external parties to see precisely how close each bank came to breaching key thresholds.
Integration with macroprudential tools: results feed into CCyB decisions (system-wide buffer) and microprudential PRA buffer decisions (bank-specific), which together shape how much capital banks must hold in normal times.

Disclosure and Reporting Practices

Effective communication of stress test results and scenarios is crucial for market confidence and for banks’ own planning. The BoE’s disclosure approach has evolved to become more transparent over time. Here’s how it compares:

In 2022 (ACS and STDF Manual): The BoE published a comprehensive results document for the ACS as part of its Financial Stability Report (FSR). For example, the July 2023 FSR published the results of the 2022/23 ACS, including charts of each bank’s CET1 ratio over the scenario and the system aggregate losses. Alongside, the BoE published the “Key elements of the 2022/23 annual cyclical scenario” which detailed the scenario assumptions (macro paths, etc.), and a “Variable paths” Excel with the exact data series. A “Traded risk scenario” Excel was also released to show the market shock parameters. Furthermore, a Guidance for participants document (likely internal or shared with banks but also made public for transparency) was provided, though such guidance is mostly of interest to the participating banks.

The public results typically include:

Aggregate system impacts: total losses, reduction in capital ratios, breakdown by credit risk, market risk, misconduct, etc.
Bank-specific outcomes: lowest CET1 ratio (and sometimes leverage ratio) reached by each bank, and comparison to its reference hurdle. Often this is presented as a chart or bar graph.
Narrative commentary on what drove each bank’s result (e.g., bank X was most affected due to its large mortgage portfolio hit by house price falls, etc.).
Confirmation of whether any bank fell below its requirements and if so, what actions were taken (in recent years, usually none fell below after factoring management actions).
The FPC/PRC’s conclusions, such as “the system is resilient” or “the test indicates banks can support the economy in a stress.”

In terms of data transparency: The BoE does not publish the raw data submissions (which are extremely granular and confidential). But by releasing the scenario data and summarizing outcomes, it allows external analysts to simulate or understand the test to a degree. The BoE also historically would give qualitative feedback letters to each participating bank privately, detailing strengths and weaknesses of their stress testing approach and results.

The STDF manual itself, while public, is more of a technical reference (89 pages of definitions and process). It’s not something the average market analyst would use, but its public availability underscores the BoE’s transparency about how it collects and handles data. It also helps other banks (not in the test) to align their internal stress test data with these definitions if they wish.

In 2024 (desk-based test): As noted, only aggregate results were published. The November 2024 FSR contained a section summarizing the 2024 stress test findings. The BoE likely commented on overall capital adequacy and maybe compared the severity of the scenario to previous ones. However, no individual bank results were provided. This was a departure from the usual practice, but understandable since those results were modelled by the BoE and not vetted by banks. The BoE did publish the scenarios themselves on 27 June 2024 (with paths for GDP, unemployment, etc.), and also a short results document on 29 November 2024. The FSR entry “2024 desk-based stress test results” likely contained charts of system aggregate CET1 ratio over the scenario, maybe a comparison of two scenario impacts, and commentary that helped justify maintaining or adjusting the CCyB.

By only disclosing aggregate outcomes, the BoE maintained confidentiality of any individual firm’s performance. This was a one-off, and the BoE signalled that in non-concurrent tests, it doesn’t plan to reveal bank-specific data. This ensures banks aren’t unfairly stigmatized by a regulatory model’s output that they didn’t have a hand in producing.

In 2025 (BCST disclosures): We expect a return to full transparency, as the BoE has explicitly stated it will publish both aggregate and bank-level results. The novelty will be showing low point vs requirements for each bank, as discussed. This will likely be presented either as a table or explained in prose in the results publication. The BoE might say, for example, “Bank A’s CET1 ratio fell to a low of X%, which is Y percentage points above its minimum requirement of Z%. Bank B’s fell to M%, which is at the level of its minimum requirement (or below by N percentage points), etc. The PRA judged that given the temporary nature of the dip and high provisions, this did not indicate capital inadequacy” – something along those lines to contextualize it.

The BoE has already published on 24 March 2025:

“Key Elements of the 2025 Bank Capital Stress Test” – describing the scenario highlights and the fact that it’s the first under the new framework.
“Variable paths for the 2025 stress test” (Excel) – all the macroeconomic and financial variable trajectories in the scenario.
“Traded risk scenario for the 2025 stress test” (Excel) – the market shock details.
“Guidance for participants – 2025 Bank Capital Stress Test” – the detailed instructions for banks (which we saw excerpts of, covering definitions and expectations).
All of these being public means any informed stakeholder can see exactly what conditions are being tested and what rules apply. This openness is aimed at strengthening credibility: market participants know the test’s severity (so they can form their own view if it’s tough enough), and banks know the playing field is level.

After banks submit their projections and BoE finalizes results (likely by Q4 2025), the BoE will publish the results (probably December 2025). Typically, this comes as a standalone report or as part of the Financial Stability Report. Given the importance of this first BCST, the BoE might publish a dedicated document. It will have the features discussed – including charts, tables, and narrative. Another disclosure aspect is model methodology: BoE often publishes box articles in the results or FSR about any particular method changes. For instance, in 2025 they might include “Box: Incorporating IFRS 9 in the stress test” (which we saw references to Box A in guidance). This would explain to the public the changes in timing and buffer usage we’ve discussed. They may also include a Box on climate or other risk if they did sensitivity analysis. The Key Elements 2025 already includes Box B about scenario calibration adjustments.

The level of detail in disclosures has to balance informativeness with not spilling sensitive data (like proprietary model details or extremely granular exposures). The BoE seems to have struck a balance: publish enough to justify policy decisions and demonstrate resilience, but not so much that, for example, a bank’s exact loss rate on a certain portfolio is revealed (which could be market-sensitive or confidential).

Historical data and references: The BoE often references previous results for comparison. In the approach document, they cite results of past tests (2022/23 ACS, 2021 Climate BES). We may see in 2025’s results something like “this test’s severity is broadly similar to the 2019 ACS in terms of peak unemployment and house price fall, and the aggregate CET1 drawdown (X percentage points) is similar to that in 2018”, etc., giving context. That helps readers calibrate continuity.

Disclosure to firms vs public: Each participating bank, besides the public info, receives private feedback/supervisory letters. By the time of results publication, those banks also know what their PRA buffer requirement will be (since the PRA usually communicates new capital requirements to them around the same time, if not earlier). The BoE has said banks wanted more private information on drivers of their results. Possibly the BoE might provide more detailed benchmarking to each bank privately (like how its loss rates compared to peers, etc.) in 2025 to address that feedback.

To encapsulate:

2014-2022: High transparency, including bank-level results vs hurdles.
2024: Only system-level transparency (anomaly due to transitional approach).
2025: Return to high transparency, with even more clarity on usage of buffers in stress. BoE is essentially as open as any major regulator on stress tests, akin to how the EBA EU-wide stress tests publish a lot of data (though BoE doesn’t publish full balance sheet data like EBA, which publishes detailed templates – BoE has not gone that far likely due to proprietary considerations and differences in approach).

Conclusion

The Bank of England’s 2025 stress testing framework represents an evolutionary step from its earlier versions, aligning the test frequency and design with the current risk environment and regulatory priorities. Relative to 2024 and 2022, the 2025 framework introduces a biennial cycle (BCST replacing the annual ACS), incorporates the end of IFRS 9 transitional arrangements, and refines scenario calibration – all while maintaining the same rigorous standards of resilience. The Annual Cyclical Scenario that defined the 2014–2022 era has given way to a Bank Capital Stress Test conducted every two years, underpinned by the Stress Test Data Framework introduced in 2022 and still in use. In tandem, the BoE has enhanced its use of exploratory scenarios to investigate non-cyclical threats, scheduling them flexibly around the BCST.

Key changes such as the timing of scenario shocks and treatment of capital buffers in 2025 ensure that the benefits of IFRS 9’s forward-looking provisioning are recognized without compromising the severity of the test.

The changes from 2024’s approach – where a one-off desk-based test was run – to 2025 illustrate the BoE’s willingness to adapt its methodology (using lighter or more agile techniques when appropriate). Yet, with the launch of the 2025 BCST, the stress testing regime returns to full strength, combining bank-generated data and supervisory scrutiny to assess resilience. Crucially, the BoE and PRA will use the 2025 stress results to inform both microprudential and macroprudential actions, from setting individual bank capital buffers to guiding the Countercyclical Capital Buffer for the system.

In terms of disclosure and accountability, the 2025 framework continues the BoE’s transparent approach, committing to publish granular results and the rationale behind them. Stakeholders will be able to compare the 2025 outcomes side-by-side with prior exercises, aided by the BoE’s clear communication of any methodological changes (e.g., Boxes on IFRS 9 in official publications). By doing so, the BoE ensures that the stress testing framework not only serves as a supervisory tool but also as a public confidence measure, demonstrating that UK banks can “absorb rather than amplify shocks” in line with the Financial Policy Committee’s objectives.

In conclusion, the 2025 BoE stress testing framework builds on the foundations of the 2022 STDF/ACS and the transitional steps seen in 2024, resulting in a more streamlined yet equally robust regime. It balances consistency (unchanged risk tolerance and data standards) with innovation (biennial cycle, flexible exploratory tests), thereby reinforcing the resilience of the UK banking system in an efficient and forward-looking manner. All these changes are carefully calibrated to maintain confidence that, even under extreme scenarios, UK banks will have the capital strength to support households and businesses – fulfilling the ultimate aim of stress testing.