Reverse stress testing explained

A board convenes to approve the annual Internal Capital Adequacy Assessment Process (ICAAP). The capital ratios are healthy. The stress scenarios have been calibrated carefully and the firm’s financial resources clear every one of them with room to spare. The chief risk officer steps through the results. There are no red flags. Then a non-executive leans forward with the question the pack has no page for: what would actually bring this firm down?

That is the question reverse stress testing exists to force. It doesn’t start with a scenario and project a capital impact. It starts with the endpoint, business-model failure, and works backward to the events that would cause it. The difference isn’t presentational. A conventional stress test tells a firm whether it can survive what its scenario designers anticipated. A reverse stress test forces the firm to find what its scenario designers did not.

A conventional stress test asks whether the firm survives the storm it saw coming. A reverse stress test goes looking for the one it didn’t.

How does reverse stress testing differ from a conventional stress test?

A bank can’t wait to discover whether it would survive a crisis by having one. Stress testing exists to answer that question in advance, and the conventional form answers it forwards: push a range of adverse scenarios through the firm’s financial model and measure the effect on capital resources. The exercise is useful for capital planning, because it quantifies the buffer a firm needs to withstand a defined set of adverse outcomes.

Reverse stress testing inverts this. Instead of deriving the capital impact of a chosen scenario, the firm fixes the outcome, business-model failure, and searches for the pathways that lead there. The Prudential Regulation Authority (PRA) describes it as "a risk management tool used to increase a firm’s awareness of its business model vulnerabilities". The practical value of this inversion is that it frees the analysis from the limits of pre-chosen scenarios. Conventional stress tests, however carefully designed, are bounded by the imagination and institutional biases of the people who chose the scenarios. A reverse stress test is bounded by the definition of failure itself.

The obligation is broad. Banks, building societies, investment firms, and insurers are all required to run the exercise, whether through Chapter 15 of the PRA Rulebook’s Internal Capital Adequacy Assessment Part or the Financial Conduct Authority’s parallel SYSC 20 requirement. The most useful document, though, is older: FG11/07, issued in 2011 by the then Financial Services Authority (FSA), remains the most detailed operational commentary on how to actually run the exercise, and it is the source practitioners still work from.

Failure before the capital runs out

The definition of failure is the hinge of the whole exercise, and the PRA is unambiguous that it doesn’t mean capital exhaustion.

“The point at which the market loses confidence in a firm and, as a result, the firm is no longer able to carry out its business activities. Such a point may be reached well before the firm’s financial resources are exhausted.”PRA, SS31/15, §4.3

Business-model failure is a confidence event, not an accounting event. The PRA frames failure as the moment at which counterparties, investors, and other stakeholders become unwilling to transact with the firm or to provide it with capital. Capital and liquidity may still exist when this happens; what has been lost is the market’s belief that the firm can continue as a going concern. Defining failure as a confidence event, not an accounting one, is also why reverse stress testing belongs to the board rather than the modelling team.

The FSA’s FG11/07 gives practitioners three entry points for developing failure scenarios, and they are three stages of one chain. The cause: market participants see the firm as over-exposed to a particularly risky sector. The consequence: counterparties refuse to deal, or will only deal on terms so onerous the business becomes unviable. The impact: the firm cannot transact new business and its revenue streams dry up. A firm may pick up the chain at whichever stage it finds most natural to reason from.

Crucially, the FSA is direct that business model failure "is not solely about inadequate financial resources". An endpoint defined in purely capital terms misreads the exercise.

The implication for practitioners is that a scenario library confined to financial shocks will miss the reputational and operational paths that can trigger the confidence event before the balance sheet shows a problem.

Where the rulebook puts it

Today the reverse stress testing requirement has its own dedicated chapter in the PRA’s supervisory statement SS31/15, having started life as a cross-reference in the general stress testing regime. The detail worth knowing is proportionality, and FG11/07 sets the test by size rather than by which supervisory exercises a firm sits in: it expects the process to be "primarily qualitative in nature for smaller firms", and "more detailed analysis, incorporating quantitative analysis from the outset" from larger and more complex ones.¹ The documentation requirement in Chapter 15 of the Internal Capital Adequacy Assessment Part applies regardless of size: firms must record the results and the mitigating measures identified.²

How do firms choose which failure scenarios to develop?

Scenario selection works by elimination. The FSA’s FG11/07 is clear that firms should start by considering a wide range of scenarios and narrow them down to the ones most likely to cause business-model failure. The filter isn’t a probability cut. Scenarios qualify as "the most likely scenarios, given that business model failure is a prerequisite", and "not necessarily based on an assessment of the absolute probability". The question isn’t how often an event like this occurs. It is whether it could credibly happen to this firm, given how this firm is built, and whether it would be fatal if it did.

The PRA offers illustrative starting points: the failure of one or more major counterparties, or market disruption caused by the failure of a major market participant. These aren’t prescribed scenarios but illustrations of the kind of event that can cause business-model failure through confidence contagion rather than direct financial loss to the firm itself. A liquidity run is the archetypal case; our companion pieces on ILAAP³ and liquidity stress testing⁴ show the same logic applied there.

Where the board comes in

Reverse stress testing isn’t a technical exercise to be delegated to the risk function with a sign-off at the end. The FSA’s FG11/07 states plainly that "the Board or senior management of the firm must sign off the reverse stress test", and that Board sign-off "is recommended in any event", providing the opportunity to use the exercise and its outputs to inform the firm’s risk appetite and business planning.

A board is doing something qualitatively different from approving a capital ratio when it engages seriously with the question of what would actually end this firm. It is testing whether the business model holds together under adverse but plausible conditions, identifying which management actions would in fact be sufficient, and building institutional awareness of the firm’s deepest vulnerabilities into strategic decisions. The results and the mitigating measures identified must be documented in the firm’s ICAAP under Chapter 15 of the Internal Capital Adequacy Assessment Part.

Boards that treat the sign-off as a rubber stamp miss the exercise’s main management benefit: a structured opportunity to challenge the firm’s assumptions about its own resilience.

What the supervisor does with it

The Bank of England runs its own capital-setting exercise, the Bank Capital Stress Test, which since 2025 has been held "every other year" rather than annually.⁶ Separately from that, the PRA uses a firm’s own reverse stress-testing results through three distinct channels.

• ICAAP mitigants. The documented results and associated management actions feed into the ICAAP, informing how the firm addresses the vulnerabilities the exercise has surfaced.
• SREP input. The results feed the PRA’s Supervisory Review and Evaluation Process (SREP). The PRA may request the design and results of the exercise (§5.4), and it can act early on what it finds, because the PRA "recognises that not every business failure is driven by lack of financial resources" (§5.9).²
• Recovery-plan seed. The scenarios a firm identifies through reverse stress testing may provide the starting point for the scenarios used in recovery planning.

The position on capital has been consistent since the FSA’s 2011 guidance: supervisors "will not use the results of the exercise to generate additional capital requirements directly".¹ Indirectly is another matter. The exercise can lead to higher requirements where it exposes weaknesses in a firm’s oversight and governance. Reverse stress testing isn’t a capital floor. It is a tool for business-model awareness and governance.

The whole flow fits in one diagram: the results at the top, their three supervisory destinations beneath.

Treated as a compliance formality, reverse stress testing produces documentation that satisfies Chapter 15 and changes nothing. The versions that work are the ones where the scenarios are genuinely uncomfortable, the board has challenged the scenario designers, and the results have altered how the firm thinks about its own fragility.

Which returns us to the boardroom. The non-executive’s question was never an interruption; it was the whole point of the meeting, waiting to be asked. A firm that has run its reverse stress test honestly walks in already knowing the answer: the scenarios that would end it, how close they sit, and what it has done about them. That is what the exercise is for. Not a number for the supervisor, but an answer for the board.

One test tells you which kind of exercise you have. Did the reverse stress test change anything: a limit, a plan, a decision? If nothing moved, it was documentation.

Frequently asked questions