RESEARCH DOSSIER
Reverse Stress Testing for Insurer Capital Failure
Abstract. Reverse stress testing begins with a defined state of failure and investigates combinations of conditions capable of producing it. For insurers, this requires a more exact definition of failure than a single fall in reported capital: economic depletion, breach of a jurisdiction-specific solvency requirement, breach of a minimum capital requirement, inability to meet payments when due, and the opening of supervisory recovery or resolution processes are related but distinct events. This monograph develops a source-grounded comparison of reverse stress testing under the International Association of Insurance Supervisors' Insurance Core Principles, the European Solvency II regime, U.S. risk-based capital arrangements, and the United Kingdom's prudential approach. It examines model formulation, market and insurance-loss dependencies, counterparty and reinsurance exposures, liquidity feedback, management interventions, scenario plausibility, model uncertainty, and board and supervisory use. An explicit illustrative balance-sheet calculation constructs multiple failure boundaries, derives a minimum-distance capital-breach scenario, and tests a coupled shock path with solvency and liquidity thresholds. The analysis demonstrates how an apparently adequate initial capital ratio can coexist with materially shorter distances to failure under combinations of asset losses, reserve strengthening, claims-payment acceleration, and contingent cash demands. It also establishes why regulatory ratios from different regimes cannot be interchanged, why a mathematical nearest point is not necessarily an economically plausible event, and why a management action that cures liquidity may leave solvency impaired. The contribution is a comparative interpretation and reproducible illustrative analysis grounded in existing standards and literature. The work proposes no new named theory, doctrine, or proprietary model.
Keywords: Reverse stress testing; insurance solvency; capital failure; economic balance sheet; Solvency II; solvency capital requirement; minimum capital requirement; risk-based capital; Insurance Capital Standard; ORSA; scenario optimization; joint shocks; reserve deterioration; market risk; reinsurance recoverables; liquidity; collateral; contagion; management actions; recovery planning; scenario plausibility; model risk; insurance supervision.
Introduction
The intellectually decisive question in insurer reverse stress testing is what conditions would cause a specific insurer to lose the capacity to honor its financial commitments or satisfy an identified prudential boundary. Conventional forward testing specifies an adverse event and evaluates its consequences; reverse testing specifies an adverse consequence and searches for causal paths to it. The distinction changes the research problem. A forward scenario may reveal an unexpected capital loss. A reverse inquiry must explain why its selected failure definition is economically, legally, and operationally meaningful, identify the factors that can cause it, and assess whether the causal combination is credible at the insurer's business horizon.
The International Association of Insurance Supervisors (IAIS, 2024) incorporates reverse stress testing into the Insurance Core Principles and ComFrame, particularly the risk-management and enterprise-risk-management provisions of ICP 16. European authorities implement own-risk-and-solvency-assessment expectations through EIOPA's ORSA Guidelines (EIOPA, 2015). The U.S. National Association of Insurance Commissioners (NAIC, 2026a) describes risk-based capital as an intervention-oriented statutory measure, rather than a comprehensive solvency determination. These different emphases are central to the research: a useful reverse exercise can illuminate failure mechanisms common to insurance businesses while measuring them against legal and actuarial boundaries that remain regime-specific.
The inquiry covers life, health, property-casualty, and composite insurance, including groups with cross-border entities and reinsurance connections. It considers six interacting risk families: investment market losses; underwriting, claims, and reserve deterioration; credit and reinsurance counterparty impairment; cash and collateral strain; operational and concentration losses; and intervention or recovery capacity. The primary unit of analysis is the insurance undertaking, although group-level transmission receives separate treatment. This scope is deliberately broader than a single one-year capital ratio because an insurer can possess substantial accounting assets and still face payments or collateral demands before those assets become available.
The research method combines direct examination of prudential texts, comparative regulatory interpretation, specialist scenario-design literature, officially reported insurance stress-test results, and a transparent deterministic numerical example. The numerical figures are hypothetical and expressed in abstract monetary units. They demonstrate the mathematics of failure-boundary reasoning; they are not estimates of any company's balance sheet, a fitted probability distribution, supervisory calibration, or an assessment of actual financial strength. Reported empirical findings are attributed to their original institutions and separated from the present author's illustrative calculations.
The organizing argument is that reverse stress testing becomes decision-relevant when four objects are defined together: the failure threshold, the sequence of shocks, the insurer's balance-sheet and cash-flow responses, and the operational feasibility of management actions. Where one of these objects remains implicit, a seemingly precise reverse stress result may answer the wrong economic question. This argument is an analytical synthesis of the cited supervisory and research literature, not a separate theoretical system.
Capital Failure Conditions
The economic balance-sheet boundary
An insurer's economic balance sheet measures the value of assets against a valuation of insurance and other liabilities. Simplifying prudential adjustments, economic net assets can be written as , where is the recognized asset value and is the value of liabilities. A reverse exercise exploring exhaustion of net assets searches for shocks such that . This boundary captures the balance-sheet mechanics of asset losses and liability increases, but it does not alone establish legal insolvency, a failure of payment liquidity, or the point at which a supervisory authority must intervene. Those conclusions depend on the valuation regime, capital eligibility rules, local law, institutional guarantees, and timing of cash flows.
Insurance liabilities are particularly sensitive to contingent obligations. A windstorm or liability-claim development may increase expected future payments before those payments become due. An interest-rate movement may simultaneously alter the discounted value of guaranteed life liabilities, bond assets, and the value and collateral requirements of derivatives used for asset-liability matching. Recoverable amounts from reinsurance or risk-transfer contracts may protect net liabilities on an economic basis while remaining exposed to delayed collection, counterparty impairment, coverage disputes, or collateral restrictions. Consequently, a credible failure definition distinguishes economic valuation from cash settlement and the legal availability of individual assets.
Solvency Capital Requirement and Minimum Capital Requirement
Under European Solvency II, Directive 2009/138/EC, Article 101 stipulates a Solvency Capital Requirement (SCR) calibrated to a one-year 99.5% value-at-risk measure of basic own funds under the specified regulatory assumptions. Its SCR calculation provisions cover insurance underwriting, market, credit, and operational risks (EIOPA, n.d.-a). A simple coverage indicator is , subject to the relevant eligibility restrictions. A 100% coverage crossing is an SCR breach in the illustrative ratio sense, while Article 138 governs notification and restoration procedures for the actual breach. The SCR noncompliance rules call for immediate notification, with a recovery plan and a prescribed restoration process subject to its detailed conditions (EIOPA, n.d.-b).
The Minimum Capital Requirement (MCR) provides a separate supervisory boundary. Under Article 139, MCR breach procedures involve a short-term realistic finance scheme and tighter restoration deadlines (EIOPA, n.d.-c). The actual MCR calculation is more complex than choosing a fixed percentage of the SCR. Delegated Regulation (EU) 2015/35 includes an MCR combined-calculation structure with a linear component, an SCR-based corridor, and an absolute floor (European Commission, 2015). A research example may assume a proportional MCR for illustration, but must disclose that assumption and never substitute it for a legal computation.
The insurer can therefore cross three capital-related boundaries at different shock intensities: loss of its SCR buffer, deterioration below the MCR threshold, and economic exhaustion of net assets. For an undertaking that begins with and , the solvency ratio is 156.25%, not the distance in monetary units to payment failure. The initial monetary headroom above the SCR is 90; its erosion depends both on movements in eligible own funds and on changes in the SCR denominator. A market stress can reduce the numerator while simultaneously changing the risk calculation, so simply subtracting investment losses from an unchanged capital requirement may materially misstate the threshold.
U.S. statutory risk-based capital boundaries
The U.S. RBC framework has different accounting conventions, risk aggregation, and legally actionable ratios. The NAIC's risk-based capital explanation (2026a) describes authorized control level RBC and progressive regulatory intervention. RBC is calculated using statutory accounting and line-specific formulas for life, property-casualty, and health insurers. A ratio expressed as total adjusted capital to authorized control level RBC is not numerically interchangeable with EU eligible own funds divided by SCR. The denominator, underlying capital concept, action levels, and legal authority are different. A reverse stress exercise for a U.S. insurer must reconstruct the relevant statutory capital calculation, trend-test provisions where applicable, and the state-level implementation of model laws rather than transport a Solvency II threshold into a U.S. template.
The NAIC's ORSA framework (2026b) adds enterprise-level prospective risk analysis and group-wide solvency considerations for covered insurers. ORSA is a crucial setting for reverse stress inquiry because a statutory ratio observed at a reporting date may fail to expose a path that develops across underwriting cycles, asset sales, policyholder behavior, and intragroup support assumptions. Forward-looking group capital adequacy is analytically distinct from any one legal entity's ability to pay claims in its own currency and jurisdiction.
International group capital and supervisory comparability
The IAIS adopted the Insurance Capital Standard (ICS) in December 2024 as a prescribed capital requirement within ComFrame for internationally active insurance groups. The IAIS ICS overview describes the international quantitative capital project (IAIS, 2024–2026). Adoption of an international standard must be distinguished from its application through local jurisdictional arrangements and from the facts of individual group supervisory determinations. The IAIS's implementation assessment principles (2025) further establish that assessment of implementation is an ongoing separate task. A cross-border reverse test should report the requirements of each relevant supervisor and capital regime, rather than assuming that one supranational group measure substitutes automatically for all solo-entity constraints.
A related timing point concerns legal evolution. Directive (EU) 2025/2 amends the Solvency II framework, with its substantial transposition and application schedule extending to January 30, 2027. As of the November 15, 2026 research cutoff, the Solvency II amending directive (European Parliament and Council, 2025a) warrants forward-looking implementation analysis, not treatment as though all amended requirements were already operative. The same caution applies to the Insurance Recovery and Resolution Directive (European Parliament and Council, 2025b), whose framework and future application must be separated from existing domestic resolution powers.
Liquidity failure, operational continuity, and policyholder payment
A fourth boundary arises when obligations become payable and the undertaking cannot obtain legally available cash quickly enough. This liquidity condition may be reached while net assets remain positive and while a solvency ratio appears compliant. Conversely, an insurer with a breached capital requirement may continue paying claims for some time under supervisory constraints. Operational failure also takes forms other than a cash shortfall: access to payment infrastructure, settlement counterparties, claims administration, and policy data can become impaired in a cyber or natural-catastrophe event even when investment assets have not suffered their ultimate valuation loss.
For analysis it is useful to distinguish four failure events: , eligible capital below the applicable SCR; , eligible capital below the applicable MCR; , available cash and executable financing below dated payments; and , nonpositive adjusted economic net assets. Their union is not a universal legal definition of insurer failure. It is a deliberately explicit collection of research boundaries that must be assigned legal meaning and supervisory consequences in the selected jurisdiction. A reverse exercise may target the earliest of these events, or report separate scenarios for each. Reporting them separately is more informative than collapsing all capital, liquidity, and resolution states into the word “insolvency.”
Reverse Stress Problem Formulation
Starting from a failure set
Let represent a vector of coherent adverse risk drivers. Depending on the insurer, these may include sovereign yields, corporate credit spreads, equity returns, inflation, claims frequency, severity, mortality, longevity, lapses, reserve development, catastrophe loss, reinsurance default, collateral haircuts, and operational downtime. Let denote contractual and regulatory constraints and let represent management actions that are feasible at the time and conditions being modeled. A forward model maps these inputs to balance-sheet valuations, capital requirements, dated cash flows, and operational service capacity:
The reverse problem specifies a failure set in the outputs of and searches for satisfying . The most basic capital breach has the equality boundary and a failure region . Liquidity has a dated path condition , where is cumulative freely available cash or realizable funding after all obligations falling due by date .
A failure set also specifies the business perimeter. A group can report a consolidated capital surplus while a subsidiary lacks transfer-accessible capital, or while a reinsurance subsidiary possesses assets that are encumbered for policyholder protection. The reverse model should calculate solo legal-entity constraints, group-level constraints, and intragroup transfers only where law, contracts, fungibility, and timing support them. Group-wide modeling without a legal-entity map can conceal the very failure mechanism that the exercise is supposed to reveal.
Selecting an objective: severity, distance, or plausibility
There is no universal mathematical meaning of the “smallest” failure scenario. Three common choices deserve explicit distinction. A minimum-distance problem finds for a stated metric , but depends on the scale and correlations used in that metric. A minimum-cost adversarial scenario finds the least damaging combination of shocks that breaches a target, conditional on a specified stress budget. A plausibility-constrained optimization searches over scenarios within a defensible empirical or judgmental region and identifies those producing severe consequences. Each is legitimate only relative to its assumptions.
Breuer, Jandačka, Rheinberger, and Summer's (2009) plausible severe scenario analysis shows how stress-test design can search for maximum loss within a plausibility domain and how the treatment of unstressed factors matters. Breuer and Csiszár's (2013) entropic plausibility constraints provide another mathematical route to stress regions. These contributions are principally mathematical or banking-oriented research; applying them to an insurer requires explicit insurance balance-sheet maps and contractual constraints. Neither publication establishes that Euclidean distance in arbitrarily scaled risk factors equals probability of occurrence.
An optimization also faces a temporal issue. A capital breach today may be reached through an asset-market move, while an adverse reserve revision emerges over several reporting periods. An insurer could suffer payment failure tomorrow because collateral cash is due immediately even though the capital loss is measured at year-end. Thus the state vector should include when and in what sequence shocks occur. Where a calculation intentionally suppresses time, the report must describe the result as a static, diagnostic boundary rather than a forecast of the sequence to insolvency.
Management actions as conditional controls
A management-action variable belongs in the model only when its execution conditions are transparent. Retaining dividends, issuing common equity, reducing underwriting, obtaining reinsurance, selling bonds, changing hedging positions, negotiating collateral terms, and transferring portfolios have different execution lags, price effects, contractual limits, and supervisory approvals. Optimizing over all conceivable management actions as though they were freely available under stress produces an artificially resilient undertaking. The opposite convention—excluding all actions—can exaggerate capital deterioration and omit an important response channel.
A rigorous comparison calculates at least three cases: fixed balance sheet, prespecified credible management actions, and constrained or failed execution of those actions. In its 2024 insurance stress test, EIOPA distinguishes fixed-balance-sheet and management-action views; this demonstrates why they should not be treated as the same reported result (EIOPA, 2024a). The benefit attributable to action should be calculated net of transaction cost, realized loss, financing expense, counterparty and legal risk, and adverse effects on remaining policyholders.
Data provenance and numerical reproducibility
A fully reproducible reverse exercise records the valuation date; accounting and prudential bases; legal-entity and group perimeter; currency; liability cash-flow projections; asset model and market data; claims triangles and reserve assumptions; reinsurance counterparty, attachment, and collateral terms; capital eligibility; scenario horizon; optimization and search constraints; management-action rules; and the exact failure definition. Model version, human overrides, and uncertainty ranges belong in the same audit record. These requirements follow from the interpretation of insurer cash flows and capital mechanics. They also correspond to the supervisory aim of enterprise risk management under IAIS ICP 16 (IAIS, 2024).
An individual model run is not a finding about population-wide insolvency probability. For that inference, the underlying model must specify a probability law, credible calibration, time dependency, and validation against relevant data. Reverse stress testing often has a different purpose: to identify mechanisms that management has not adequately considered, define actionable limits, and challenge seemingly comfortable initial capital positions. A numerical point on a failure surface can be precise arithmetic while its likelihood remains unknown.
Joint Market and Insurance Shocks
Why single-factor shocks can conceal capital fragility
The capital consequence of a single market-factor movement depends on the insurer's assets, liabilities, hedges, reinsurance, and underwriting composition. A bond portfolio's spread widening can lower market values and simultaneously make liquidity less accessible; if recession also produces higher unemployment, lapse rates and health claims may change. A property catastrophe may increase gross claims at precisely the time an insurer's reinsurers and investment portfolio experience adverse market conditions. A long-tail liability reserve increase can coincide with social inflation and greater uncertainty in the distribution of ultimate settlement costs. An insurer's vulnerability is determined by the combined balance-sheet effect, the covariance and nonlinear interaction of drivers, and the legal timing of cash flows—not by the sum of independently selected, unrealistic maxima.
The conditional structure matters. Equity losses and corporate spread widening may share a recession driver; catastrophe losses and reinsurance disputes may share contract and event characteristics; margin calls and cash sales may form a feedback loop. Independence cannot be assumed merely because a capital calculation is modular. The analyst should identify which dependencies follow economically from a common cause, which are empirically estimated, and which remain stress assumptions. An optimizer that combines each separate worst event without a coherent common narrative may create a logical impossibility. An optimizer that omits dependencies may miss a plausible multi-risk collapse.
Insurance underwriting requires line-specific pathways. For a life insurer, falling yields may increase the value of long-duration guarantees, while rapid yield increases can induce surrenders, liquidity demands, and losses on bonds sold before maturity. For property-casualty companies, unmodeled catastrophe concentration and delayed casualty reserve development may dominate capital loss, particularly where reinsurance protection is limited by reinstatement premium, exhaustion of aggregate limits, or an unsettled coverage dispute. Health insurers face claims utilization, medical cost trends, provider failure, and regulatory pricing constraints. Composite groups may possess offsetting risks but the timing and fungibility of those offsets must be demonstrated.
A transparent hypothetical insurer
Consider a synthetic insurer at the valuation date. Every number below is an illustrative amount in the same unspecified monetary unit, and every coefficient is assumed solely for explaining the method. Initial assets are and liabilities are . Assume that all of the resulting simplified net assets qualify as eligible own funds; that is an explicit simplifying assumption, not a representation of Solvency II eligibility. The baseline SCR is . Hence the starting SCR coverage is $250/160=156.25%$ and the absolute SCR headroom is .
The stress vector comprises normalized equity or investment-market deterioration , credit spread deterioration , adverse reserve development , counterparty impairment , and forced-sale price stress , with $0\leq e,c,r,q,f\leq1$. Normalization is arbitrary in this example. Adopt the following linear, illustrative maps:
The coefficients do not claim to reproduce a regulator's capital formula, an empirical insurance dataset, or any known inter-risk correlation matrix. They are intentionally disclosed stress-response sensitivities. They combine both sides of the solvency ratio: a market or claims shock can destroy own funds and increase the modeled capital charge. Accordingly, the threshold for an SCR breach is
For a given shock combination, and should be recomputed from the same risk vector. Treating an adverse change in the denominator as an independent second loss, or implicitly double-counting the same deterioration in assets, leads to an incorrect capital deficit. The reserve coefficient here records a change in liability valuation; the separate illustrative reserve capital loading reflects greater remaining underwriting uncertainty, rather than the same known claim settlement entered twice.
Minimum-norm breach and its limitations
Suppose a mathematical exercise seeks the smallest Euclidean shock vector in the normalized coordinates that brings the insurer to the SCR boundary. Let . For an interior solution to , minimization of yields . Since , the solution is
This solution respects all imposed factor bounds and produces zero SCR headroom up to rounding. It is mathematically reproducible by substitution. Its interpretation is narrow: it is the nearest breach under one arbitrarily normalized Euclidean norm, with no probability model and no enforced macroeconomic or insurance dependency constraints. If the reserve axis is rescaled, or correlation and temporal structure are introduced, the point moves. A responsible reverse test therefore reports the assumptions behind the optimization criterion and challenges the economic coherence of the resulting combination.
Breuer and colleagues' (2009) scenario plausibility research is relevant here precisely because severity and plausibility are distinct choices. A fitted likelihood region or entropic stress neighborhood, such as that examined in Breuer and Csiszár (2013), could replace the Euclidean objective if appropriate actuarial inputs exist. Such a change would produce a different mathematical problem, and its solution would inherit the uncertainties of the fitted distribution.
A coupled stress path through the failure surfaces
A second calculation applies a coherent illustrative progression along one prechosen risk direction: for $0\leq t\leq1t$ represents scenario intensity, not calendar time or frequency. Inserting this path into the equations gives
The SCR boundary is reached when . At that severity the insurer still has positive own funds, but the numerator is just equal to the capital requirement. That result illustrates a threshold crossing, not inevitable bankruptcy at that instant. By full scenario intensity, , own funds are , the modeled SCR is , the coverage ratio is $39.77%132.5$.
To illustrate a lower regulatory capital boundary without reproducing all legal MCR mechanics, assume solely for this example that the MCR remains $45%$ of the modeled SCR across the scenario path, with the relevant absolute floor below that amount. Then , and the MCR crossing satisfies $250-162.5t=72+27t$, or
At , the toy MCR is and the eligible-own-funds shortfall against it is . The order of crossings in this constructed model is SCR at followed by MCR at . In a real EU calculation, technical provisions, risk modules, the linear MCR, the prescribed corridor, absolute floors, and capital eligibility must be recomputed under the applicable rules, including the formal MCR calculation provisions. The example is therefore a demonstration of threshold analysis, not a regulatory filing.
The same shock vector reaches economic exhaustion of own funds at , beyond the modeled interval. Thus neither the full stress nor the MCR breach implies economic net assets are mathematically zero in this example. This distinction is essential: using “capital failure” to mean one specific prudential breach is analytically permissible, but it should not smuggle in an assertion that accounting net worth has vanished or every claim has become unpayable.
Relevance of empirical European stress evidence
EIOPA's 2024 insurance stress results show how a large forward exercise can reveal capital sensitivity while preserving distinct interpretations of solvency and liquidity (EIOPA, 2024a). The reported aggregate starting ratio of 221.8% and a scenario capital decline exceeding €270 billion concern the participant sample and specified prescribed shocks, not the hypothetical insurer above. EIOPA expressly described the exercise as a supervisory diagnostic rather than a simple pass/fail test. The corresponding full 2024 report (EIOPA, 2024b) provides methodological detail and fixed-versus-constrained-balance-sheet distinctions. Translating those observations into a reverse stress project involves solving for each insurer's own failure boundaries; it does not mean extrapolating the reported sample averages into individual capital requirements.
Liquidity and Contagion Feedback
Dated obligations and executable funding
The liquidity problem requires a dated ledger of legally available sources and contractually due uses, separated by currency and legal entity. Sources may include free cash, maturing investments, securities capable of sale or repo, contractual premium income, collectible reinsurance recoverables, and committed facilities. Uses may include scheduled policyholder claims, accelerated catastrophe claims, derivative variation margin, securities collateral, surrender payments, tax, operating expense, and redemption or debt-service commitments. Asset saleability depends on market depth, settlement time, spread, haircuts, encumbrance, and the insurer's ability to transfer proceeds across subsidiaries. A reverse liquidity stress should therefore calculate for each significant interval, not simply compare the current balance-sheet total with ultimate liabilities.
The International Monetary Fund's 2024 financial-stability work on the Netherlands discusses derivatives, variation margin, and insurer liquidity risk in the Netherlands FSAP analysis (IMF, 2024a). That setting demonstrates why interest-rate hedges can protect economic value while creating demands for cash collateral. It also illustrates the difference between assessing solvency and mapping the path of liquidity resources; model results must remain specific to the observed sector and its institutional environment. The IMF's Luxembourg stress-testing assessment (IMF, 2024b) similarly requires attention to static assumptions, group and investment exposures, and excluded behavioral reactions before transferring any country-specific numerical outcome to another insurer.
A reproducible liquidity crossing
Extend the illustrative insurer by defining a single-horizon liquidity resource envelope. At baseline, assume cash of 50, securities executable at the selected horizon for 135, and ordinary non-reinsurance contractual inflows of 20, totaling 205. A separate recognized reinsurance receivable discussed below is excluded from these executable resources pending collection. These 135 securities form part of the assets already included in the balance sheet; they are not newly created financial capital. In the adverse path , assume executable funding is reduced by $27t$ because of stress haircuts and timing constraints. Assume baseline cash uses of 90 and incremental cash requirements as follows: catastrophe or reserve-related claim acceleration $60r(t)=54t, margin calls \40e(t)=32t, surrenders \30c(t)=21t, and reinsurance payment delays or related funding needs \15q(t)=9t. The total incremental uses are \116t$.
The illustrative liquidity headroom is therefore
Liquidity becomes inadequate at . By , the funding gap is . Along this hypothetical direction the SCR limit is reached first, the liquidity limit second, and the MCR limit third. Different initial buffers, margining terms, or haircut schedules could reverse the liquidity/MCR ordering. The result is thus a comparative diagnostic about a specified synthetic undertaking rather than a universal ordering of insurance failure events.
The liquidity calculation assumes that the securities are executable within the horizon and that incoming amounts are collectible. A stress test must validate these conditions against market settlement, account controls, policyholder asset rings, security lending terms, collateral eligibility, and the possibility of payment holidays or disputed reinsurance claims. If the 135 of assumed realizable securities cannot be monetized in time, the liquidity crossing moves toward a less severe shock. Conversely, contractual committed facilities may increase usable resources, but their drawdown conditions and lender performance during the same crisis must be tested.
Feedback from asset sales and funding stress
The model initially treats forced-sale stress as an independently entered component of the scenario. In a realistic mechanism it may be endogenous. A margin call raises the liquidity deficit; the deficit causes liquidation of securities; sales at stressed prices reduce own funds; lower capital coverage may prompt additional counterparties to demand collateral or raise financing costs. A self-consistent model can represent this with a joint fixed point, for example , where denotes funding stress. The feedback must be solved jointly rather than counting an assumed shock and an independently modeled sale loss for the same transaction twice.
A reverse simulation that propagates prices and balance-sheet decisions across multiple institutions needs assumptions about common asset holdings, collateral contracts, market impact, and secondary sales. Baes and Schaanning's (2023) reverse macroprudential scenario design explores how reverse stress problems and balance-sheet feedback can be analyzed in a financial-system setting. The study's banking and macroprudential methods provide intellectual tools for thinking about endogenous deleveraging; an insurer application must replace banking funding and default mechanisms with actual insurance liabilities, asset duration, reinsurance, and policyholder behavior. The present monograph does not infer empirically measured insurance contagion magnitudes from that banking research.
Reinsurance as capital relief and conditional cash support
A reinsurance contract creates multiple economic links that require separate treatment. It can reduce a cedant's net exposure and capital charge, but the cedant may remain obligated to pay policyholders before the reinsurance recovery arrives. Catastrophic events can exhaust reinstatement limits, generate disputed aggregate attachment calculations, and create industry-wide counterparty demands. An insurer concentrated in one reinsurer or group may experience a correlation between high claims and the financial weakness of its reinsurer. Some contracts require pre-funding, collateral, or reserve credit conditions that change under stress.
Consequently, a reverse stress analysis should record gross claims, ceded recoverables, timing of payment to claimants, collectible counterparty payments, and whether the capital charge genuinely incorporates counterparty credit risk. The SCR risk-mitigation rules recognize risk mitigation while requiring corresponding credit and other risks to be reflected (EIOPA, n.d.-a). A scenario that simply subtracts gross reinsurance recoverables from gross claims without considering liquidity, credit, or coverage terms will overstate short-horizon resilience.
Group contagion and entity-level restrictions
Cross-border insurer groups may transmit stress through capital calls, centralized treasury, asset management mandates, reinsurance pools, common derivative counterparties, intra-group guarantees, and governance decisions. The same structure can also diversify earnings and provide access to capital. The relevant inquiry is which transfers remain legally permitted, financially possible, and operationally executable after each breach. A parent group's consolidated surplus may not satisfy a subsidiary's local collateral or claims-payment requirement. Foreign-exchange controls, local policyholder protection measures, and capital-transfer approvals can turn nominally fungible group resources into inaccessible support at the point of need.
The IAIS ComFrame group-supervision provisions (IAIS, 2024) supply the international supervisory context for considering such linkages. A practical reverse exercise should show group and major solo boundaries simultaneously, report the legal hierarchy of claims and collateral, and specify whether the modeled parent's support is a signed enforceable commitment or a discretionary intention. Where a reverse scenario identifies a concentrated group vulnerability, it is the causal mechanism and feasible control that matter; a single consolidated stress ratio can be too coarse to guide action.
Scenario Plausibility and Uncertainty
Conditional plausibility and risk-factor distributions
A reverse scenario can be mathematically feasible without being economically credible. Plausibility asks whether the combination of movements could arise under the specified market, underwriting, and institutional mechanisms. It is not equivalent to assigning a frequency to every scenario. Some scenarios can be tested because their mechanism is strategically critical and physically or legally possible even when empirical data are too sparse to support a stable tail probability. The analyst should distinguish a conditional severe scenario (“if a major cyber event coincides with market stress”) from an unconditional empirical return period (“once in 200 years”), which requires a defensible joint distribution and time horizon.
Breuer et al. (2009) explain in their stress scenario methodology that the treatment of unshocked factors affects the plausibility of a constructed event. If an analyst conditions on a recession, nonselected market factors should not automatically be held at normal values where conditional movements are economically required. This insight is directly relevant to insurers with duration-sensitive liabilities, joint lapse behavior, and spread-sensitive asset values. Breuer and Csiszár (2013) examine entropy-constrained stress design, but actuarial application requires careful calibration of the distribution and suitable treatment of low-frequency catastrophe and litigation risk.
Historical data offer essential constraints but only partial information about tail dependence. Market crises are observable in financial time series, whereas long-tail insurance reserve deterioration may emerge after years and catastrophe exposure changes with climate, inflation, construction, and underwriting composition. Extreme cyber events also involve rapidly changing interdependencies and relatively short industry data histories. Hence a documented expert stress may complement estimated distributions; its causal assumptions should be explicit and open to challenge. Frequency claims must be supported by the underlying data rather than inferred from the intensity parameter in a synthetic scenario path.
Multidimensional uncertainty and model risk
Parameter uncertainty affects reserve development, correlations, catastrophe secondary perils, derivative netting, reinsurance collection, liquidity haircuts, and behavioral reactions. Model uncertainty concerns the structural validity of chosen equations: whether the insurer's capital requirements rise linearly with shocks, whether policyholders change behavior, whether asset markets remain liquid, and whether recovery options survive systemic stress. Legal uncertainty concerns contract enforceability, supervisory waivers, capital transfer restrictions, and time to authorization. Operational uncertainty concerns available staff, payment systems, data integrity, and execution during simultaneous operational disruption.
These uncertainty classes suggest sensitivity testing around the assumptions that determine the failure boundary, rather than merely varying an arbitrary shock scalar. In the illustrative capital equation, increasing the reserve-loss coefficient from 50 to 70 without changing the reserve component of the SCR causes the capital threshold to be reached earlier along the stated path. The altered reserve contribution raises the combined erosion coefficient from 222.5 to 240.5 because the path includes , so instead of . This shows how apparently modest reserve-model changes can matter when the initial headroom is small relative to the combined shock response. The alternative case is an assumption test, not a calibrated actuarial confidence interval.
A further robustness check can make the cash-haircut coefficient more adverse, change the availability date for reinsurance recoveries, and make the management response conditional on financial-market depth. Each perturbation should identify which decision changes: earliest capital breach, liquidity emergency, MCR breach, or the feasibility of a planned remedy. A general-purpose sensitivity chart that changes coefficients without recording which failure boundary or governance decision moves is less informative than a mechanism-focused audit.
Evidence from official insurer stress exercises
EIOPA's 2024 EU insurance exercise reports capital deterioration under a prescribed adverse geopolitical and macroeconomic scenario while emphasizing the sector's resilience in the selected sample (EIOPA, 2024a). The report explains that initial capital resources were substantial, describes fixed versus management-action variants, and distinguishes cash holdings from wider liquid-asset holdings. These observations matter methodologically because they show that large market-driven capital reductions do not automatically imply payment failure, and that liquidity assessments require usable assets and dated cash flows. They cannot be reinterpreted as a historical finding that no EU insurer could fail under a more severe or different scenario.
The Bank of England's UK insurer stress-testing overview (Bank of England, n.d.) documents the supervisory role of insurer stress exercises. Its 2025 life insurance stress results (PRA, 2025) show how a UK-specific life insurance sample responds to scenario and management assumptions; they are not a universal benchmark for general insurers or U.S. property-casualty companies. The 2026 general insurance stress update (PRA, 2025–2026) illustrates the importance of exercise design and reporting dates. A study prepared on November 15, 2026 must not present results scheduled for a later release as though they already existed.
Legal date, jurisdiction, and comparability discipline
A regulatory assertion requires its jurisdiction, instrument, and temporal applicability. The EIOPA rulebook records Solvency II Directive text and relevant implementing measures; national law determines implementation and supervisory administration. The UK prudential approach, documented in PRA Supervisory Statement SS19/16, is a separate regime with its own amendment schedule (PRA, 2016–2026). The U.S. framework uses state-adopted statutory rules and NAIC model laws. IAIS standards express international supervision expectations and must be mapped through application arrangements before being treated as binding company-level requirements.
The date problem is substantive, not merely bibliographic. At the research cutoff, the extensive EU 2025 reform legislation has a prospective 2027 implementation milestone. The EU amended Solvency II text and insurance recovery legislation need future-regime labels where the provisions are not yet effective. Comparisons should therefore distinguish prevailing minimum requirements from planned resolution architecture. No universal 99.5% one-year rule can be imputed to every insurance regulatory capital system simply because it is specified for EU SCR.
Research validity and limits
The numerical exercise is deterministic, linear, and deliberately parsimonious. Its assumptions are fully disclosed and mathematically checkable, but its coefficients lack statistical estimation, its stress axes have arbitrary units, and its management-action and liquidity rules simplify real market mechanics. It neither computes default probabilities nor assigns a jurisdiction-specific risk-based capital filing to a real insurer. Its importance is logical and diagnostic: by holding one set of assumptions fixed, the reader can trace the threshold sequence and see exactly how a change in assumptions shifts it. Actual applications require validated firm data, actuarial reserve assessments, contract-level cash-flow and reinsurance detail, realistic risk-factor dependence, and supervisory review appropriate to the jurisdiction.
The empirical literature and regulatory reports have their own boundaries. Banking stress optimization contributes mathematical tools but not insurer-specific default frequencies. Supervisory sample stress tests reveal resilience under selected scenarios but are not universal evidence of inability to fail. Legislation establishes regulatory obligations but does not provide the balance-sheet data needed to compute an individual company's threshold. A sound synthesis preserves these distinctions rather than assigning every source the same evidentiary role.
Recovery and Governance Use
The governance decision starts at the failure boundary
A reverse stress analysis is useful when its results can be translated into decisions about capital, liquidity, underwriting, reinsurance, and operational continuity. The board and senior management first need to determine which outcome is intolerable, over what horizon, and for which legal entity. They then need the shortest credible causal narratives reaching that state and an assessment of whether those narratives overlap with existing risk limits. The exercise can expose a portfolio concentration that conventional risk appetite metrics obscure, a reinsurance payment dependency that dominates short-term cash flow, or a capital-management option that is legally unavailable once the regulatory trigger has been crossed.
The IAIS enterprise risk management requirements (IAIS, 2024) integrate stress testing within the governance of risk appetite and prospective solvency assessment. The EIOPA ORSA guidelines (EIOPA, 2015) place own-risk assessment within the undertaking's risk processes and recognize methods including reverse stress testing. These standards support a research sequence in which the failure definition is board-owned, the scenario search is actuarially and financially disciplined, and the proposed response is tested against execution constraints.
A board decision should not consist of declaring an acceptable probability of failure when the reverse stress method has not estimated such a probability. More defensible outputs are identifiable vulnerabilities; contingent funding gaps; capital-headroom sensitivity; early-warning indicators; feasible preventative changes; and the dated authority responsible for initiating a specific action. These outputs can inform quantitative risk limits without pretending that every scenario has an empirically validated frequency.
Recovery actions must be tested in the same scenario
For the synthetic insurer at , own funds are , SCR is , and the illustrative MCR is . Liquidity headroom is . Consider a fully disclosed management response: new capital of that is subscribed, paid, legally available, and recognized as eligible own funds; a portfolio action that realizes an additional loss; a genuine reduction of modeled SCR by through risk reduction; and collection of an additional of reinsurance cash that was already recognized as a receivable but excluded from the stressed executable-liquidity resources, assumed for this toy calculation to convert that asset into cash without increasing economic net assets. This receipt is separate from the incremental short-horizon funding need already counted in the stress-flow coefficients.
Under these assumptions, post-action eligible own funds are $87.5+30-8=109.5$ and the revised SCR is $220-20=200. The solvency ratio becomes \109.5/200=54.75%. The insurer is still far below its SCR, even though the action increases its own funds above the example's post-shock MCR of \0.45\times200=90508$ sale loss is treated as a reduction in net executable sale proceeds, leaving in liquidity headroom. This demonstrates that restoration of immediate payment capacity does not imply restoration of SCR coverage, and that solvency improvement through newly recognized capital differs from the mere collection of an already recognized asset.
The assumed transaction costs, SCR decrease, and MCR proportionality are pedagogical. An actual transaction could change the denominator, eligibility, reserves, taxes, and collateral demands differently; a capital injection may be delayed, legally restricted, or unavailable in a systemic crisis. A credible reverse stress report would present this post-action case alongside failed issuance, delayed reinsurance, and more severe haircut variants. It would also state whether capital raising and asset sales can occur before legal intervention, because a strategy that can only be executed after contractual or supervisory restrictions arise cannot be credited without qualification.
Prioritizing prevention over last-minute repair
Preventive actions alter the failure surface before a crisis. They may include reducing concentrated asset exposures, lowering duration or currency mismatch, diversifying reinsurance recoverables, securing collateral terms with appropriate liquidity reserves, revising gross and net catastrophe limits, tightening claims reserve monitoring, strengthening contingency funding sources, and creating documented operational continuity arrangements. Their effects are heterogeneous. More reinsurance may reduce underwriting loss while increasing counterparty or timing risk; shifting into cash may improve liquidity while reducing expected return or changing asset-liability mismatch. Prevention should therefore be assessed on both solvency and liquidity dimensions under the same scenario assumptions.
The nearest mathematical capital breach can guide a search for weak points, but governance should emphasize causal narratives. For example, if reserve deterioration consistently appears in the lowest-severity coherent failure paths, stronger reserve diagnostics and line-level claims controls may be more effective than changing an unrelated investment limit. If liquidity failure occurs only when collateral demands coincide with reinsurance delays, a funding facility or altered contractual settlement terms may be more efficient than holding a much larger undifferentiated capital buffer. These are conditional implications of model results, not unqualified prescriptions for insurers generally.
Supervisory escalation and capital triggers
A reverse stress exercise needs a trigger map aligned with applicable law. In the EU, notification and restoration obligations differ between SCR and MCR breaches under Solvency II Articles 138 and 139, with separate MCR-specific escalation. In the United States, the relevant legal actions follow state-adopted RBC rules and statutory accounting, as outlined in the NAIC intervention framework. In an internationally active group, group-wide capital measures under the ICS standard-setting framework sit alongside the entity-level rules adopted by relevant jurisdictions. A capital trigger should therefore identify the metric, denominator, observation date, legal entity, governing authority, and escalation obligation.
Recovery and resolution planning must also recognize future legal change. The European Union's insurance recovery and resolution framework was enacted in 2025 with an implementation schedule after the current research cutoff. Institutions should analyze future obligations separately from present-day powers. The general governance lesson remains: reverse stress scenarios become actionable when early indicators are positioned ahead of irreversible actions, with sufficient lead time for realistic governance and legal approvals.
An evidence-based reverse stress report
A comprehensive report can be organized around the following audit sequence without introducing a proprietary framework: record the decision-maker and undertaking perimeter; specify each failure event and its legal basis; record the baseline balance sheet and regulatory denominator; map risk drivers and their dependencies; identify both nearest mathematical and economically coherent severe paths; simulate cash and capital through time; compute threshold crossings; test management actions and their failure modes; disclose uncertainty and sensitivity; and connect the findings to board limits, supervisory notice requirements, and operational continuity arrangements. This sequence is a practical synthesis of the preceding standards and methods.
The report must preserve the difference between a verified fact, a model assumption, and an analytical inference. A statutory trigger is supported by the controlling legal instrument. A historical sector result is attributed to the published stress-test population. A scenario coefficient is labeled as estimated, judgmental, or hypothetical. An inference about governance is supported by the demonstrated mechanism. Such provenance prevents a mathematical illustration from appearing to be an empirically observed failure and allows a later reader to reproduce, challenge, or update the work without reconstructing the author's intentions.
Comparative synthesis and institutional consequences
The comparison reveals common analytical foundations and distinct regulatory expressions. Under the IAIS standards, reverse stress testing supports enterprise risk management and forward-looking assessment. Under Solvency II, eligible-own-funds coverage, SCR and MCR are legally structured and accompanied by different escalation pathways. U.S. RBC uses statutory balance-sheet concepts, industry-specific formulas, and model-law action levels whose authority depends on domestic law. The UK's prudential and ORSA practice reflects its own supervisory statements and scenario exercises. The ICS creates an international group-capital reference whose implementation must be established at the jurisdictional level. Each regime can accommodate a search for plausible failure mechanisms, but each requires a different legal map to interpret the resulting crossing.
A second conclusion concerns time. Capital depletion, loss of payment liquidity, and operational interruption can occur in different orders. In the constructed adverse path, the SCR crossing precedes the liquidity crossing, which precedes the MCR crossing. In a concentrated catastrophe with rapid settlement obligations and delayed reinsurance, immediate liquidity may become the first binding constraint. In a reserve deterioration discovered before claims fall due, regulatory capital may bind first. A reverse stress report without a temporal claims-and-collateral ledger can therefore misidentify the earliest consequential state of failure.
A third conclusion concerns adaptation. An insurer may be able to sell assets, raise capital, seek more reinsurance, or shrink underwriting in ordinary conditions. These options tend to become harder when several insurers or financial institutions seek the same protection under the same shock. The EIOPA 2024 fixed-balance-sheet comparison and the mathematical literature on feedback-aware scenario design demonstrate why actions should be modeled conditionally rather than credited as automatic offsets. The calculation of prevented loss must include execution feasibility as well as arithmetic benefit.
Finally, the most valuable outcome of reverse stress inquiry may be a change in the insurer's understanding of what constitutes survival. A high reported coverage ratio can coexist with concentrated contingent cash obligations, inaccessible group resources, or a dependency on a single reinsurer. Conversely, a stressed ratio below a supervisory threshold does not by itself establish that all claim payments will stop immediately. Explicit definitions, transparent mathematics, institutional context, and action-specific governance allow these facts to coexist without conceptual confusion.
Conclusion
Reverse stress testing for insurer capital failure is a disciplined search backward from a precisely specified adverse state to the combinations of market, underwriting, counterparty, liquidity, and operational conditions capable of causing it. Its scientific usefulness depends on accurate failure-boundary definitions, validated valuation and cash-flow mechanisms, defensible shock dependencies, and governance assumptions that survive the modeled conditions. The method is fundamentally comparative: different failure surfaces reveal different vulnerabilities and lead to different corrective decisions.
The supervisory evidence establishes why an insurer capital-failure threshold must be assigned to a particular jurisdiction and legal entity. Solvency II's SCR and MCR have distinct requirements and procedures; U.S. risk-based capital relies on statutory metrics and progressive state-law intervention; IAIS capital and enterprise-risk-management standards operate through international supervisory architecture and jurisdictional implementation. These metrics are not interchangeable. Planned legislative changes, including major EU reforms with future application dates, must be described in their proper temporal status.
The illustrative analysis provides reproducible numerical conclusions under explicitly hypothetical assumptions. With starting eligible own funds of 250 and SCR of 160, the minimum Euclidean normalized stress reaching an SCR breach is $90a/20{,}150$ for . Along the separate illustrative adverse path, the SCR threshold is reached at severity , the liquidity threshold at , and the assumed proportional MCR threshold at . Full-severity own funds remain positive at 87.5 despite severe regulatory capital impairment, demonstrating that capital-ratio breach and economic net-worth exhaustion are different boundary conditions. A conditional recovery case restores short-horizon liquidity headroom while leaving SCR coverage at 54.75%, demonstrating that a liquidity remedy can coexist with continued capital failure. These values are analytical consequences of the stated coefficients and provide no empirical probability or insurer-specific forecast.
The literature on plausibility-constrained stress design and balance-sheet feedback supplies useful methodological tools. Their transfer to insurance requires actuarial claims and reserve dynamics, reinsurance contracts, legal-entity restrictions, collateral timing, and conditional management behavior. The central governance outcome is a causal record linking threshold, scenario, feasible action, and regulatory response. Original scholarly synthesis here lies in joining these elements across established supervisory regimes and in making the distinctions testable through an explicit calculation. No new named theory, conceptual doctrine, or unverified priority claim is introduced.
References
Baes, M., & Schaanning, E. (2023). Reverse stress testing: Scenario design for macroprudential stress tests. Mathematical Finance, 33(2), 209–256. https://onlinelibrary.wiley.com/doi/10.1111/mafi.12373
Bank of England. (n.d.). Stress testing insurers. Bank of England. https://www.bankofengland.co.uk/stress-testing/insurers
Breuer, T., & Csiszár, I. (2013). Systematic stress tests with entropic plausibility constraints. Journal of Banking & Finance, 37(5), 1552–1559. https://www.sciencedirect.com/science/article/pii/S0378426612001112
Breuer, T., Jandačka, M., Rheinberger, K., & Summer, M. (2009). How to find plausible, severe, and useful stress scenarios. International Journal of Central Banking, 5(3), 205–224. https://www.ijcb.org/journal/v5n3/how-find-plausible-severe-and-useful-stress-scenarios
European Commission. (2015). Commission Delegated Regulation (EU) 2015/35, Article 248: Calculation of the Minimum Capital Requirement. EIOPA Solvency II Single Rulebook. https://www.eiopa.europa.eu/rulebook/solvency-ii-single-rulebook/article-5908_en
European Insurance and Occupational Pensions Authority. (2015). Guidelines on own risk and solvency assessment (ORSA). EIOPA. https://www.eiopa.europa.eu/system/files/2022-10/eiopa_guidelines_on_orsa_en.pdf
European Insurance and Occupational Pensions Authority. (2024a). Insurance stress test 2024: Results and recommendations. EIOPA. https://www.eiopa.europa.eu/insurance-stress-test-2024_en
European Insurance and Occupational Pensions Authority. (2024b). Insurance stress test 2024: Report. EIOPA. https://www.eiopa.europa.eu/document/download/f8a234b0-a84a-49ff-975e-c47f8849bfc0_en?filename=Report+-+Insurance+Stress+Test+2024.pdf
European Insurance and Occupational Pensions Authority. (n.d.-a). Solvency II Single Rulebook: Directive 2009/138/EC, Article 101, calculation of the Solvency Capital Requirement. EIOPA. https://www.eiopa.europa.eu/rulebook/solvency-ii-single-rulebook/article-2188_en
European Insurance and Occupational Pensions Authority. (n.d.-b). Solvency II Single Rulebook: Directive 2009/138/EC, Article 138, noncompliance with the Solvency Capital Requirement. EIOPA. https://www.eiopa.europa.eu/rulebook/solvency-ii-single-rulebook/article-2225_en
European Insurance and Occupational Pensions Authority. (n.d.-c). Solvency II Single Rulebook: Directive 2009/138/EC, Article 139, noncompliance with the Minimum Capital Requirement. EIOPA. https://www.eiopa.europa.eu/rulebook/solvency-ii-single-rulebook/article-2226_en
European Parliament and Council. (2025a). Directive (EU) 2025/2 amending Directive 2009/138/EC as regards proportionality, quality of supervision, reporting, long-term guarantee measures, macroprudential tools, sustainability risks, and group and cross-border supervision. Official Journal of the European Union. https://eur-lex.europa.eu/eli/dir/2025/2/oj/eng
European Parliament and Council. (2025b). Directive (EU) 2025/1 establishing a framework for the recovery and resolution of insurance and reinsurance undertakings. EUR-Lex legislative summary. https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=LEGISSUM%3A5027527
International Association of Insurance Supervisors. (2024). Insurance Core Principles and Common Framework for the Supervision of Internationally Active Insurance Groups (ComFrame): December 2024. IAIS. https://www.iais.org/uploads/2025/01/IAIS-ICPs-and-ComFrame-adopted-in-December-2024.pdf
International Association of Insurance Supervisors. (2024–2026). Insurance Capital Standard (ICS). IAIS institutional overview. https://www.iais.org/activities-topics/standard-setting/insurance-capital-standard/
International Association of Insurance Supervisors. (2025). High-level principles for the development of the ICS implementation assessment published. IAIS, July. https://www.iais.org/2025/07/high-level-principles-for-the-development-of-the-ics-implementation-assessment-published/
International Monetary Fund. (2024a). Kingdom of the Netherlands—The Netherlands: Financial Sector Assessment Program—Technical note on systemic risk analysis. IMF Country Report No. 2024/174. https://www.elibrary.imf.org/view/journals/002/2024/174/article-A001-en.xml
International Monetary Fund. (2024b). Luxembourg: Financial Sector Assessment Program—Technical note on stress testing and systemic risk analysis. IMF Country Report No. 2024/185. https://www.elibrary.imf.org/view/journals/002/2024/185/article-A001-en.xml
National Association of Insurance Commissioners. (2026a). Risk-based capital. Last updated June 30, 2026. NAIC. https://content.naic.org/insurance-topics/risk-based-capital
National Association of Insurance Commissioners. (2026b). Own risk and solvency assessment. NAIC. https://content.naic.org/insurance-topics/own-risk-and-solvency-assessment
Prudential Regulation Authority. (2016–2026). Solvency II: Own risk and solvency assessment (SS19/16). Bank of England. https://www.bankofengland.co.uk/prudential-regulation/publication/2016/solvency2-orsa
Prudential Regulation Authority. (2025). Life Insurance Stress Test 2025: Results report. Bank of England. https://www.bankofengland.co.uk/prudential-regulation/publication/2025/november/list-2025-results-report
Prudential Regulation Authority. (2025–2026). Update on 2026 dynamic general insurance stress test (DyGIST). Bank of England. https://www.bankofengland.co.uk/prudential-regulation/publication/2025/july/update-on-2026-dygist