Enterprise risk management connects governance, risk measurement and financial decisions. This guide develops those connections through 20 original practice questions and 60 concepts covering risk ownership, emerging risks, dependence, stress testing, hedging, reinsurance, ALM and capital allocation. Calculations state their assumptions, while examples distinguish economic value, accounting results and timely cash. Scope follows the supplied November 2026/March 2027/July 2027 syllabus for the next November administration, viewed as of October 16, 2026. The questions use MCQs as a learning format; the supplied evidence confirms an examination case study but does not establish the official response method. Source s3 supports syllabus relevance; technical explanations use established subject knowledge.
What this guide covers
Explore CFE 101 ERM through 20 original practice questions, worked reasoning, and 60 technical concepts.
Enterprise Risk Management Foundations
1. ERM value through financial frictions
ERM coordinates risks across an enterprise so decisions reflect their combined effect on objectives. Its economic value often arises through financial frictions: distress costs, costly external financing, taxes, customer confidence and constraints on investment. Reducing volatility is useful when it improves these outcomes enough to justify mitigation costs. A hedge that merely removes a diversifiable fluctuation need not create shareholder value. Assess ERM by comparing feasible enterprise outcomes with and without the proposed framework, including implementation expense, altered incentives and opportunities to undertake profitable business.
Worked example: An insurer spends USD 1 million annually on liquidity protection. If that protection reduces expected forced-sale and emergency-financing costs by USD 1.6 million, the estimated annual net benefit is USD 0.6 million before other effects.
Common mistake: Lower reported earnings volatility does not by itself establish value creation; the economic benefits must exceed mitigation and implementation costs.
Official syllabus reference: [1]
2. COSO ERM links strategy and performance
COSO ERM connects governance and culture, strategy and objective-setting, performance, review and revision, and information, communication and reporting. Its mechanism is integration: risk considerations influence which strategy is selected, how objectives are set and how performance is interpreted. A risk register alone does not implement this connection. Application requires examining alternative strategies, identifying risks to their execution and judging whether the resulting profile fits appetite. Review then revises assumptions and responses as performance or external conditions change, rather than treating the framework as a one-time compliance exercise.
Worked example: Before entering a new distribution channel, an insurer compares expected sales, conduct exposure and operational capacity. It sets growth objectives conditional on complaint indicators and service capacity, then revises expansion when those assumptions deteriorate.
Common mistake: Completing framework documentation does not demonstrate integration if strategy selection and performance decisions continue without using the identified risks.
Official syllabus reference: [1]
3. Three lines preserve ownership and assurance
A three-lines arrangement distinguishes management of activities, specialist support and challenge, and independent assurance. Business management owns risks and operates controls; risk and compliance functions provide expertise, monitoring and challenge; internal audit assesses governance and control effectiveness independently. The governing body oversees accountability across the arrangement. Collaboration should share information without merging responsibilities that require independence. Apply the model by naming decision owners, challenge authorities and assurance responsibilities for important processes. Outsourcing a service or obtaining risk-function approval does not remove the business owner's accountability for the resulting exposure.
Worked example: For outsourced claims handling, the claims director owns service risk, compliance tests customer-treatment requirements, and internal audit assesses whether outsourcing oversight works. A shared incident log supports collaboration while each function retains its distinct responsibility.
Common mistake: Risk-function involvement does not transfer operating ownership away from management, and internal audit should not become the routine approver of business decisions.
Official syllabus reference: [1]
4. CRO authority and functional independence
An effective risk function needs sufficient expertise, access to information, resources and authority to challenge material decisions. Functional independence means that commercial pressure cannot silently suppress risk conclusions; it does not require isolation from the business. Reporting routes to senior management and the board should permit timely escalation of unresolved concerns. The structure should match enterprise complexity, combining central standards with specialist knowledge where needed. Evaluate effectiveness through the quality of challenge, follow-through and decision influence, rather than through headcount or the number of risk reports produced.
Worked example: A subsidiary proposes a concentrated investment. Its local risk officer identifies a group-wide exposure that local reports miss, and the group CRO can escalate the unresolved limit issue directly to the board risk committee before approval.
Common mistake: A senior title alone does not create independence when information access, escalation rights and compensation incentives still depend entirely on commercial management.
Official syllabus reference: [1]
5. Principal-agent incentives and deferred rewards
Agency problems arise when a decision-maker's incentives differ from those of the party bearing the consequences. Short measurement periods, private information and limited downside participation can encourage excessive risk or underinvestment in controls. Contracts, monitoring, deferred compensation and balanced performance measures can reduce these conflicts, but each introduces cost and potential gaming. Application begins by identifying who decides, who benefits and who absorbs losses across time. Measures should reward durable outcomes that managers can influence while avoiding reliance on a single proxy that obscures customer harm or long-tail exposure.
Worked example: A sales executive earns bonuses immediately, while poor persistency emerges later. Deferring part of the award and linking it to verified persistency and conduct outcomes makes future consequences relevant to today's sales decisions without relying on volume alone.
Common mistake: Replacing one narrow bonus metric with another does not resolve agency risk if managers can manipulate the replacement or shift losses beyond the measurement period.
Official syllabus reference: [1]
6. Adverse selection versus moral hazard
Adverse selection concerns hidden information affecting who enters a transaction or which terms they accept. Moral hazard concerns behavior after an arrangement changes incentives or limits another party's ability to observe actions. Both involve asymmetric information, but their timing and remedies differ. Screening and differentiated contract terms can address selection; monitoring, deductibles and incentive-compatible obligations can address behavior. Neither distinction implies misconduct in every case. In ERM, examine insurance customers, distributors, vendors and internal managers to determine whether the concern is private risk information, subsequent actions or a combination.
Worked example: A vendor knows before bidding that its systems have severe weaknesses, creating selection concerns. After winning a fixed-fee contract, it reduces maintenance effort because the insurer bears outage costs, creating a separate moral-hazard problem requiring different controls.
Common mistake: Treating every information problem as moral hazard misses pre-contract selection mechanisms and can lead to monitoring that never addresses the original hidden information.
Official syllabus reference: [1]
7. Preventive, detective and corrective controls
Preventive controls reduce the chance that an undesirable event occurs, detective controls identify events or exceptions, and corrective controls restore the process or limit consequences. A sound design combines these functions around a clearly stated failure mechanism. Segregation of duties is preventive, reconciliation is detective, and recovery procedures are corrective. Controls can fail through poor design, weak execution, override or shared dependencies. Assessment must therefore distinguish design adequacy from operating effectiveness and evaluate residual exposure after realistic control performance, rather than assuming a documented control always works.
Worked example: To reduce unauthorized payments, an insurer separates beneficiary maintenance from payment approval, reconciles approved changes to bank instructions daily, and maintains a rapid recall procedure. Each measure addresses a different stage of the same payment-risk mechanism.
Common mistake: A detective reconciliation does not prevent the initial payment error, although timely detection can reduce its duration and eventual financial consequences.
Official syllabus reference: [1]
8. Conduct risk follows customer outcomes
Conduct risk concerns behavior and business practices that produce inappropriate customer or market outcomes. It can arise through product design, distribution incentives, communications, claims handling and service failures, even when individual transactions satisfy narrow procedural checks. Assess the customer journey and identify where information imbalance or incentives can create harm. Complaints, cancellations, claim reversals and distribution patterns provide evidence, but low complaints may reflect barriers to reporting. Relevant obligations depend on jurisdiction and product; an ERM assessment should avoid substituting an invented universal rule for the applicable requirements.
Worked example: A product passes disclosure checks, but older customers repeatedly cancel after discovering surrender charges they did not understand. The insurer examines sales explanations, distributor incentives and customer testing rather than treating signed disclosure forms as conclusive evidence of suitability.
Common mistake: Low complaint volume does not prove good conduct when customers lack information, face reporting barriers or have not yet experienced the product's limitations.
Official syllabus reference: [1]
9. AI governance separates accuracy from accountability
AI governance assigns responsibility for model purpose, data use, deployment, changes and consequences. Predictive accuracy is only one dimension: biased outcomes, privacy exposure, security weaknesses, unreliable explanations and uncontrolled updates can remain material. Controls should reflect the decision's impact and include validation, access restrictions, version records, human review where meaningful and rollback capability. A human reviewer is effective only with sufficient information, authority and time to challenge outputs. Vendor ownership of a model does not remove the deploying insurer's need to understand limitations and manage resulting business risks.
Worked example: An insurer uses AI to prioritize claim reviews. It tests error rates across relevant groups, records model versions and investigates unusual override patterns. Reviewers can reverse recommendations and pause deployment when a vendor update changes decision behavior unexpectedly.
Common mistake: Adding a human approval click does not create meaningful oversight if the reviewer cannot understand, challenge or reverse the model's recommendation.
Official syllabus reference: [1]
10. ORSA joins strategy, risk and solvency
An own risk and solvency assessment evaluates whether an insurer's resources are adequate for its particular risks and plans, including prospective changes. It connects strategy, risk appetite, capital, liquidity and management actions rather than merely reproducing a regulatory capital calculation. Scenarios should reflect material exposures and interactions, with assumptions about growth, funding and mitigation made explicit. Governance should show how results influence decisions. Specific reporting and approval requirements vary by jurisdiction, so the underlying ERM purpose should be distinguished from local legal implementation and from any one prescribed capital formula.
Worked example: An insurer projects strong premium growth but finds that a surrender shock during expansion exhausts liquid assets before its capital ratio breaches the internal target. Its assessment therefore leads to funding changes as well as capital planning.
Common mistake: Passing today's regulatory capital test does not establish prospective solvency or liquidity adequacy under the insurer's own strategy and risk profile.
Official syllabus reference: [1]
11. Risk capacity, appetite and tolerance
Risk capacity describes the amount and type of risk an organization can bear given financial resources and other constraints. Risk appetite expresses the risks it is willing to take in pursuing objectives. Tolerances and limits translate that preference into operational boundaries, although terminology varies among frameworks. Appetite should sit within capacity and include qualitative considerations where numeric measurement is weak. Application requires consistent definitions, a specified horizon and explicit treatment of stress conditions. A stated appetite becomes useful when business plans, authorities and escalation decisions are actually tested against it.
Worked example: An insurer can withstand a larger loss than management wishes to accept. The board sets a lower earnings-risk appetite and delegates business-unit limits that include a buffer for uncertainty, avoiding routine use of the full financial capacity.
Common mistake: Risk appetite is not automatically the maximum survivable loss; capacity and willingness answer different questions and can legitimately differ.
Official syllabus reference: [1]
12. KRI calibration uses action lead time
A key risk indicator should track a material exposure, a driver of loss or deterioration in control performance. Calibration connects warning thresholds to risk appetite, operating capacity, variability and the time needed to intervene. Leading indicators can support prevention, while lagging indicators verify realized outcomes. Useful reporting includes trend, threshold status, ownership and required action. Historical percentiles can inform thresholds but do not establish acceptable risk on their own. Test whether an indicator changes early enough to influence the underlying process and whether false alarms would make escalation impractical.
Worked example: A claims team needs three weeks to add trained staff. Its backlog warning level is therefore set to trigger before service capacity is exhausted, using projected arrivals and processing rates rather than waiting for missed customer deadlines.
Common mistake: A threshold chosen solely because it was rarely breached historically may provide little protection when intervention takes longer than the remaining warning period.
Official syllabus reference: [1]
13. PESTLE converts external trends into exposures
PESTLE structures external scanning across political, economic, social, technological, legal and environmental factors. Its value comes from translating a trend into a causal pathway affecting enterprise objectives, rather than compiling a list of headlines. A trend may cross categories, and categories do not establish probabilities or financial impacts. Application identifies exposed products, assets, operations and counterparties, then develops monitoring or scenarios. Distinguish the external development from the organization's vulnerability: the same interest-rate, demographic or technology change can create an opportunity for one business and a threat for another.
Worked example: A shift toward remote work is initially a social and technological trend. An insurer traces it to commercial-property borrower revenues, office demand, cyber access arrangements and employee retention, producing several distinct exposures rather than one generic external-risk entry.
Common mistake: Assigning a development to a PESTLE category does not complete risk analysis; the enterprise-specific transmission mechanism still needs to be established.
Official syllabus reference: [1]
14. Strategic risk tests business-model assumptions
Strategic risk arises when choices, execution or changes in the environment undermine the organization's objectives and business model. Analysis should identify the assumptions supporting value creation: customer demand, distribution access, pricing power, capabilities and funding. Test alternative developments that would invalidate those assumptions, including successful competitor responses. Operational metrics may reveal execution problems but can miss a strategy that is efficiently delivering the wrong offering. Application combines scenarios, strategic indicators and decision triggers so management can adapt the business model before accounting losses fully reveal the deterioration.
Worked example: An insurer's growth plan assumes continued access to one comparison platform. A platform change favoring different product features threatens acquisition economics even though internal processing remains efficient. Management tests alternative distribution arrangements and their effects on margins.
Common mistake: Strong operational performance does not establish strategic soundness when the assumptions supporting customer demand, distribution or competitive advantage have changed.
Official syllabus reference: [1]
15. Regulatory change creates implementation risk
Regulatory risk includes uncertainty about changes in requirements, interpretations, supervisory expectations and enforcement, together with the organization's ability to respond. The mechanism may affect capital, product economics, disclosures, systems or market access. Assessment separates an emerging proposal from an enacted requirement and identifies relevant entities and effective dates. A change inventory should connect obligations to accountable owners, implementation dependencies and evidence of completion. Cross-border groups must examine differences among jurisdictions rather than assuming one rule applies everywhere. Scenario analysis can evaluate uncertain proposals without presenting them as existing law.
Worked example: A proposed reporting change may require new policy-level data. Before adoption is certain, an insurer assesses data availability and system lead times, then stages preparatory work so it can respond without treating the proposal as an enacted obligation.
Common mistake: A regulatory proposal should not be described as current law, and a group policy does not automatically satisfy every subsidiary's local requirements.
Official syllabus reference: [1]
16. Rating-agency capital is a separate constraint
Rating agencies assess financial strength using capital analysis alongside business profile, earnings, risk management, liquidity and other factors. Their capital measures can differ from regulatory and internal economic measures in valuation, stresses and diversification recognition. A target rating may therefore constrain business plans even when regulatory requirements are met. Application compares the binding resource requirements under each relevant perspective and examines downgrade consequences such as collateral calls or distribution effects. Capital strength supports a rating assessment, but no single ratio universally guarantees a rating or captures the full agency judgment.
Worked example: A proposed acquisition satisfies regulatory capital requirements but increases concentration under a relevant agency's approach. The insurer assesses additional capital needs and the consequences of missing its rating objective before relying on the acquisition's projected return.
Common mistake: Meeting an internal capital ratio does not guarantee a particular rating because agency assessments include different capital methods and qualitative judgments.
Official syllabus reference: [1]
17. Funding and market liquidity reinforce each other
Funding liquidity is the ability to meet obligations when due. Market liquidity is the ability to transact in assets promptly without an excessive price concession. An institution may have positive net worth but inadequate timely cash, especially when assets are difficult to sell. Funding pressure can force sales, discounts can erode capital, and declining confidence can create further outflows. Application examines cash-flow timing, settlement, collateral eligibility and reliable funding access under stress. Asset value alone does not establish liquidity, and a committed facility still requires assessment of conditions and operational accessibility.
Worked example: An insurer faces large surrenders while private-credit assets remain valuable but slow to sell. It compares payment deadlines with cash, maturing assets and usable facilities, recognizing that emergency sales could weaken both liquidity and economic surplus.
Common mistake: Positive economic surplus does not imply sufficient cash at every payment date, particularly when asset sales require time or material discounts.
Official syllabus reference: [1]
18. Climate physical and transition transmission
Physical climate risks transmit through acute hazards and longer-term environmental changes that affect mortality, morbidity, property, operations and investments. Transition risks arise from policy, technology, market and behavioral adjustments toward different economic structures. These channels can interact and need not produce smooth annual changes. Application maps hazards and transition assumptions to specific exposures, time horizons and financial effects. ESG analysis is broader than climate and can include social and governance mechanisms. A scenario is an exploration of consequences under assumptions, not an assertion that its pathway or timing is certain.
Worked example: A life insurer examines heat-related mortality and disruption to medical services alongside reduced credit quality at carbon-intensive issuers. It uses separate transmission assumptions, then tests whether both channels could affect claims and investment resources during the same period.
Common mistake: Classifying every climate-related asset loss as physical risk overlooks losses caused by policy, technology or changing market preferences during transition.
Official syllabus reference: [1]
19. Cyber frameworks connect prevention and recovery
Cyber risk affects confidentiality, integrity and availability of information and services. A framework such as NIST's Cybersecurity Framework organizes governance, identification, protection, detection, response and recovery so defensive measures connect with business consequences. The framework is a management structure, not proof that attacks cannot occur. Application identifies critical services, privileged access, data dependencies and external providers, then tests whether incidents can be detected and contained and services restored. Recovery assumptions should include corrupted data and unavailable vendors, since backups alone do not establish that restoration will be timely or trustworthy.
Worked example: An insurer maps policy administration to its identity provider and backup service. A recovery exercise tests compromised credentials and altered records, revealing that restoring servers is insufficient unless access control and data integrity are also re-established.
Common mistake: A successful backup job does not prove recoverability when restoration speed, data integrity and dependencies on compromised systems have not been tested.
Official syllabus reference: [1]
Risk Analysis and Evaluation
20. Risk taxonomy separates causes, events and impacts
A useful taxonomy organizes exposures consistently while distinguishing causes, risk events and consequences. Weak access management is a cause, unauthorized payment is an event, and financial loss or customer harm is an impact. Mixing these levels creates duplicate entries and obscures ownership. Categories such as market, credit, insurance and operational risk support aggregation, but boundaries should be documented where events span categories. Application maps each material exposure to its source, affected objectives and responsible owner, then records relationships rather than forcing every complex event into a single isolated box.
Worked example: A ransomware incident interrupts claims processing and causes remediation expense and reputational damage. The insurer records the cyber event and its several impacts together, avoiding separate entries that would count the same remediation loss multiple times.
Common mistake: A taxonomy does not require treating connected impacts as independent risks; doing so can duplicate losses and conceal their common cause.
Official syllabus reference: [1]
21. Emerging-risk identification uses weak signals
Emerging risks are new, changing or newly recognized exposures whose characteristics may be poorly understood. Identification combines horizon scanning, expert challenge, external events, near misses and changes in business practices. A sparse loss record does not establish low exposure because consequences may develop slowly or remain undiscovered. Application describes a plausible mechanism, exposed activities, uncertainty and possible indicators before assigning precise probabilities. Separate signal credibility from potential severity: a weak signal can warrant investigation when consequences are substantial, without being presented as proof that a particular loss will occur.
Worked example: Repeated vendor release notes mention unexplained model behavior, although no claims errors have been confirmed. The insurer treats this as a signal to examine deployment controls, decision samples and common-provider exposure rather than estimating risk solely from recorded losses.
Common mistake: An emerging risk need not have a reliable loss frequency before it deserves structured investigation and proportionate monitoring.
Official syllabus reference: [1]
22. Monitoring frequency follows velocity and latency
Risk velocity describes how quickly an exposure can develop into a material consequence, while latency concerns delays before consequences become observable. Monitoring should reflect both, together with the lead time required for intervention. Fast-moving liquidity risk may require intraday observation; slowly developing conduct harm may need cohort analysis over longer periods. A dashboard's frequency is not evidence of usefulness if its inputs arrive late. Application specifies indicators, data availability, review frequency and action authority, then checks whether the full detection-to-response process can act before the relevant consequence becomes unavoidable.
Worked example: Collateral calls can arrive within hours, so a monthly liquidity dashboard cannot support timely funding decisions. Conversely, an annual lapse cohort may reveal persistent mis-selling patterns that a daily total cancellation count would obscure through noise.
Common mistake: Reporting more frequently does not create earlier warning when the underlying information is delayed or the escalation process cannot respond in time.
Official syllabus reference: [1]
23. Balance-sheet and income effects use consistent bases
Risk analysis traces a shock through assets, liabilities, equity and income using a stated valuation and recognition basis. Economic surplus changes by the asset-value change minus the liability-value change. Reported earnings can differ because accounting classifications, recognition timing, tax and presentation affect where changes appear. A cash outflow also need not equal an expense if it settles an existing liability. Application starts with a reconciled opening position, specifies the shock and follows each transmission without double counting. Economic, accounting and cash-flow analyses should be connected while preserving their different purposes.
Worked example: Assets fall by USD 12 million and economic liabilities fall by USD 4 million, so surplus falls by USD 8 million. Determining reported profit requires additional recognition rules; it cannot be inferred automatically from that economic-surplus movement.
Common mistake: A payment, an expense and an economic loss are not interchangeable; their timing and balance-sheet effects depend on the transaction and measurement basis.
Official syllabus reference: [1]
24. VaR identifies a quantile rather than tail severity
For a loss variable, VaR at confidence level α is commonly defined as the smallest loss threshold whose cumulative probability is at least α. It identifies a distributional boundary, not the largest possible loss or the average loss beyond that boundary. Its interpretation requires a horizon, valuation basis and loss definition. Discrete distributions make the quantile convention especially important. Application sorts outcomes, accumulates probability and selects the stated threshold. VaR can remain unchanged even when extremely adverse outcomes become much larger, so tail analysis should accompany it when severity beyond the threshold matters.
Worked example: Annual losses are zero with probability 95%, USD 10 million with probability 4% and USD 50 million with probability 1%. Under the lower-quantile definition, 95% VaR is zero even though substantial losses remain possible.
Common mistake: A 95% VaR of zero does not mean there is no loss risk; it describes a quantile and leaves the remaining tail unspecified.
Official syllabus reference: [1]
25. TVaR averages a fixed probability tail
TVaR, also called expected shortfall under the corresponding convention, averages losses in the worst specified probability tail. A general definition integrates quantiles from confidence level α to one and divides by 1 − α. For continuous distributions this often equals conditional expected loss beyond VaR. For discrete distributions, fractional mass at the boundary may be needed, so conditioning only on losses strictly above VaR can give a different result. Application orders outcomes from worst downward and includes exactly the required probability mass, preserving probability weights rather than averaging distinct loss amounts equally.
Worked example: Suppose losses of USD 10 million and USD 50 million occur with probabilities 4% and 1%, with zero otherwise. The worst 5% average is (0.04 × 10 + 0.01 × 50)/0.05, or USD 18 million.
Common mistake: Averaging the listed tail loss amounts equally ignores their probabilities, and conditioning above VaR can mishandle discrete probability mass at the boundary.
Official syllabus reference: [1]
26. Subadditivity distinguishes coherent tail measures
Subadditivity requires the measured risk of a combined portfolio to be no greater than the sum of separately measured risks. It formalizes a diversification principle and is one property of coherent risk measures, alongside monotonicity, translation invariance and positive homogeneity. VaR can violate subadditivity, particularly for distributions with rare losses near the confidence boundary. Expected shortfall satisfies coherence under standard integrability and convention assumptions. Application checks whether the chosen measure produces incentives consistent with the organization's aggregation decisions; subadditivity does not mean every combination produces a strictly positive diversification benefit.
Worked example: Two independent exposures each lose USD 100 with probability 4%, otherwise zero. Each has 95% VaR of zero, but their sum has 95% VaR of USD 100 because the probability of no loss is only 92.16%.
Common mistake: Diversification does not guarantee that VaR decreases relative to standalone sums; the quantile's boundary behavior can produce the opposite result.
Official syllabus reference: [1]
27. Covariance aggregation measures linear dependence
For two loss components, aggregate variance equals the sum of variances plus twice their covariance. Writing covariance as correlation times the standard deviations gives σ² = σ1² + σ2² + 2ρσ1σ2. This formula determines variance whenever the relevant second moments exist; it does not require normality. Converting standard deviation into a tail capital measure requires further distributional assumptions. Application checks units, correlation estimates and exposure consistency, then compares aggregate risk with standalone amounts. A low average correlation can still understate joint extreme losses if dependence changes in stress.
Worked example: With standard deviations of USD 6 million and USD 8 million and correlation 0.25, aggregate variance is 36 + 64 + 24 = 124. Standard deviation is approximately USD 11.14 million, below the USD 14 million standalone sum.
Common mistake: The covariance formula does not make aggregate losses normally distributed or automatically validate a normal-distribution multiplier for extreme quantile capital.
Official syllabus reference: [1]
28. Copulas separate marginals from tail dependence
A copula links marginal distributions into a joint distribution, allowing separate modeling of individual risks and their dependence. Models with similar ordinary correlations can imply very different probabilities of joint extreme outcomes. For example, a nondegenerate Gaussian copula has no asymptotic tail dependence, while a Student-t copula can exhibit joint tail dependence. Model selection should reflect the mechanisms, data and stress behavior relevant to the portfolio. Application compares aggregate tail measures across credible dependence specifications and treats sparse joint-tail observations as a source of uncertainty rather than proof of independence.
Worked example: Credit and market losses have similar fitted correlations under two models. A Student-t dependence specification produces more simultaneous severe losses than a Gaussian specification, increasing aggregate tail capital even though each marginal loss distribution remains unchanged.
Common mistake: Matching a correlation matrix does not establish that two dependence models generate the same joint extremes or the same diversification benefit.
Official syllabus reference: [1]
29. Accumulation analysis reveals common dependencies
Risk accumulation occurs when multiple exposures can be affected by the same event, location, counterparty, technology or contractual feature. Nominally different business units may therefore be less diversified than their reporting structure suggests. Analysis maps concentrations through shared dependencies and estimates combined consequences under common-cause scenarios. Exposure data should connect policies, assets, vendors and collateral where relevant. Application distinguishes true diversification from fragmentation of reporting and tests whether mitigation arrangements depend on the same stressed party or infrastructure. Accumulation can be material even when individual exposure limits are each satisfied.
Worked example: Three subsidiaries use different administrators, but all administrators rely on one cloud region. A regional outage can interrupt all three services together, so vendor-name diversification does not eliminate the underlying infrastructure concentration across the group.
Common mistake: Different contractual counterparties do not guarantee independent exposure when they rely on the same infrastructure, geography or underlying financial resource.
Official syllabus reference: [1]
30. Compounding risks create nonlinear cascades
Compounding risks interact so that the combined consequence exceeds what separate assessments suggest. A market loss can trigger collateral calls, forced sales and further price declines, while an operational outage can delay the response. Analysis should map causal sequences, feedback loops and timing instead of simply adding isolated shocks. Distinguish dependence, which describes joint occurrence, from amplification, which changes consequences after events interact. Application uses linked scenarios or dynamic models to test bottlenecks, intervention points and management actions, including whether a proposed response worsens another part of the enterprise's position.
Worked example: Bond prices fall, collateral calls consume cash and forced asset sales realize further discounts. A simultaneous payment-system outage delays access to a facility, magnifying losses beyond the sum of a standalone market shock and standalone outage estimate.
Common mistake: Adding independently estimated scenario losses can miss feedback loops because one event may change the severity, timing or feasibility of responses to another.
Official syllabus reference: [1]
31. Sensitivity, stress and scenario testing differ
Sensitivity testing varies selected inputs to understand model response. Stress testing examines severe adverse conditions, while scenario analysis develops a coherent set of conditions and transmission mechanisms. These methods overlap, but their purposes should be explicit. A scenario's internal consistency matters: interest rates, credit spreads, behavior and liquidity may need joint assumptions. Application specifies severity, horizon, valuation basis and management actions, then traces impacts through capital and cash. A severe scenario need not have an assigned probability, and a collection of individually adverse shocks is not automatically an economically coherent narrative.
Worked example: A rate sensitivity changes yields by a specified amount while holding other inputs fixed. A recession scenario additionally changes defaults, spreads and customer behavior, allowing the insurer to examine interacting consequences rather than attributing everything to rates.
Common mistake: A scenario is not automatically credible because every input is adverse; the assumed relationships and management responses must also be internally consistent.
Official syllabus reference: [1]
32. Reverse stress testing starts with failure
Reverse stress testing defines an unacceptable outcome and searches for combinations of conditions that could produce it. The outcome might be cash exhaustion, loss of market access or breach of a specified capital constraint. This differs from choosing shocks first and observing their effects. Application identifies several failure pathways, examines plausibility and determines early indicators or preventive actions. The result is not necessarily a probability estimate or a single uniquely correct scenario. Its value lies in exposing vulnerabilities that ordinary selected stresses miss, including dependence on management actions that may be unavailable.
Worked example: An insurer defines failure as inability to meet cash obligations over five days. Working backward, it identifies a combination of surrenders, margin calls and unavailable collateral that causes failure even without exhausting economic capital over that period.
Common mistake: Finding a mathematically possible failure combination is only the start; useful reverse stress testing also examines its mechanism, plausibility and actionable warning signals.
Official syllabus reference: [1]
33. Model risk includes specification and misuse
Model risk arises when a model's design, estimation, implementation or use produces misleading decisions. Specification risk concerns an unsuitable structure; parameter risk concerns estimated values; implementation risk concerns coding or operational errors; misuse concerns applying outputs beyond their valid purpose. These sources require different responses. More data may improve parameter estimates without fixing an omitted mechanism. Application documents intended use, assumptions, limitations and material decision sensitivities, then applies proportionate validation and controls. Model complexity can improve representation but also increase opacity, implementation burden and dependence on assumptions that are difficult to test.
Worked example: A capital model accurately implements its formulas but assumes surrender behavior is unchanged during market stress. Additional simulation runs reduce Monte Carlo noise while leaving the omitted behavioral response, and therefore the key specification risk, intact.
Common mistake: A larger simulation count improves numerical precision but does not repair an incorrect model structure or justify using the model for a different decision.
Official syllabus reference: [1]
34. Validation triangulates evidence
Model validation evaluates conceptual soundness, data, implementation and observed performance relative to intended use. Backtesting compares predictions with outcomes, benchmarking compares alternative methods, and sensitivity analysis identifies influential assumptions. No single test establishes universal validity, especially for rare tail events with limited observations. Application uses independent challenge proportionate to materiality, reconciles results to known cases and records unresolved limitations. Passing a recent backtest does not demonstrate resilience to structural change. Validation should inform permitted uses, overlays or remediation, rather than functioning solely as a binary approval detached from decision consequences.
Worked example: An insurer validates a mortality model by reviewing exposure data, reproducing sample calculations, comparing an alternative specification and analyzing experience by cohort. Good aggregate fit does not prevent it from identifying material errors in a newly acquired portfolio.
Common mistake: A successful aggregate backtest can conceal segment errors, compensating biases and vulnerabilities to conditions absent from the historical observation period.
Official syllabus reference: [1]
35. Real-world and risk-neutral scenario generators
Economic scenario generators produce joint paths for financial variables such as rates, returns and spreads. Real-world calibration aims to represent actual outcome probabilities for planning and risk measurement. Risk-neutral calibration supports pricing consistent with traded instruments under suitable discounting and no-arbitrage conditions. The two measures answer different questions, so pricing probabilities should not be interpreted as actual event frequencies. Application selects the measure before calibration, validates cross-variable behavior and checks time-step effects. Neither historical fit nor current price fit alone guarantees suitability for every product, horizon or risk-management decision.
Worked example: An insurer uses real-world scenarios to estimate the probability of a capital shortfall. It uses a separately calibrated risk-neutral model to value an equity guarantee consistently with option prices, preserving the distinction between forecasting and pricing.
Common mistake: Risk-neutral default or return probabilities are pricing constructs and should not be substituted directly for actual-frequency estimates in business planning.
Official syllabus reference: [1]
36. Operational loss models need event structure
A frequency-severity model represents the number of operational events and the loss per event, then combines them into aggregate loss. A Poisson frequency model may be useful when its assumptions are reasonable, but clustering or changing exposure can require alternatives. Severity data may be truncated, inconsistently recorded or dominated by rare events. Application combines internal data, relevant external evidence and structured scenarios, adjusting for scale and control changes. Qualitative causal analysis remains essential because a statistical fit can miss process redesign, common-cause failures and low-frequency events absent from the dataset.
Worked example: An insurer models routine processing errors using event counts and loss sizes, but handles a major outsourced-platform failure through an additional scenario. It does not infer the platform's tail exposure solely from frequent small internal errors.
Common mistake: A good fit to routine losses does not establish that the model captures rare common-cause operational events or changes in the control environment.
Official syllabus reference: [1]
37. Liquidity ladders track cumulative cash gaps
A liquidity ladder allocates cash inflows and outflows to time buckets and calculates cumulative usable resources. Timing, settlement, currency, collateral restrictions and legal-entity location can matter as much as total amounts. Stress assumptions should include behavioral outflows, margin calls, reduced asset-sale proceeds and impaired funding access. Application first calculates cash gaps before management actions, then adds only feasible actions with their delays and constraints. A positive balance at the end of a month does not prevent an earlier payment failure, so bucket granularity should reflect the speed of material obligations.
Worked example: An insurer begins with USD 5 million, owes USD 12 million tomorrow and receives USD 10 million three days later. Its eventual positive balance does not remove tomorrow's USD 7 million gap, which requires timely funding.
Common mistake: Netting cash flows across a broad reporting period can hide an earlier shortfall that later receipts cannot retrospectively cure.
Official syllabus reference: [1]
38. Insurance systemic risk uses transmission channels
Systemic risk concerns disruption that impairs the broader financial system or economy, rather than the isolated failure of one firm. Relevant insurance transmission channels include common asset holdings, forced sales, interconnected counterparties, collateral arrangements and withdrawal-sensitive activities. Liability structure and business activities matter, so conclusions should not treat every insurer or product identically. Application maps direct exposures and indirect feedback through markets, then tests simultaneous industry responses. A firm may appear able to sell assets individually while the sector cannot collectively do so at unchanged prices during widespread stress.
Worked example: Many insurers plan to meet stress outflows by selling the same liquid bonds. If all sell together, prices may fall and collateral needs may rise, weakening the assumptions supporting each firm's individually credible liquidity plan.
Common mistake: An institution-level stress result does not capture systemic resilience when it assumes markets absorb the same management action from every institution without price effects.
Official syllabus reference: [1]
Embedding ERM into Decision-Making
39. Appetite cascades into usable limits
Risk appetite becomes operational when enterprise preferences are translated into business limits, authorities and escalation rules. The translation should connect each limit to the relevant risk driver and recognize aggregation, diversification uncertainty and resources reserved for emerging risks. Unit limits cannot always be added mechanically because enterprise constraints may be nonlinear. Application tests proposed plans against both local limits and aggregate appetite, specifies warning buffers and assigns authority for exceptions. A breach is information requiring a defined response; silently revising the limit to fit current exposure undermines the framework's decision value.
Worked example: A group sets an enterprise credit-concentration appetite and assigns issuer limits across investment desks. A central exposure system combines holdings, so each desk's compliance does not conceal a group-level breach involving the same corporate borrower.
Common mistake: Compliance with every local limit does not guarantee enterprise compliance when exposures share counterparties or aggregate through nonlinear risk measures.
Official syllabus reference: [1]
40. Risk responses compare residual exposure and cost
Avoidance removes an activity or exposure, acceptance retains it deliberately, reduction changes its probability or impact without transferring it, and transfer shifts specified consequences to another party. Each response changes opportunity, cost and residual risk differently. Transfer can introduce counterparty, basis and legal uncertainty; avoidance can sacrifice profitable business. Application compares enterprise outcomes after response costs and secondary effects, not merely the gross risk reduction. Acceptance can be rational when exposure fits appetite and mitigation is uneconomic, provided ownership and resources are explicit rather than the exposure being overlooked.
Worked example: For a low-frequency service interruption, an insurer compares redundant systems, insurance and deliberate retention. Insurance reimburses some financial loss but does not restore service, so the preferred package depends on customer-impact tolerance as well as expected claims recovery.
Common mistake: Purchasing insurance does not necessarily transfer every consequence of an event, especially service disruption, reputational damage or losses outside contractual coverage.
Official syllabus reference: [1]
41. Incremental capital supports project selection
A project changes enterprise risk through its interaction with the existing portfolio. Incremental capital is aggregate capital after adding the project minus aggregate capital before it, rather than the project's standalone requirement. Compare expected risk-adjusted benefits with financing, mitigation and capital costs using a consistent horizon. Application also checks hard limits and liquidity, because a favorable return cannot override an infeasible constraint. For mutually exclusive projects, return ratios and value creation can rank alternatives differently; the chosen criterion should reflect the decision, resource scarcity and whether projects can actually be scaled.
Worked example: A project earns USD 2.4 million annually and adds USD 6 million to enterprise capital. At a 15% capital cost, its annual incremental EVA is USD 1.5 million, assuming the profit measure is before that charge.
Common mistake: Charging standalone capital to a proposed project can misstate its enterprise contribution when the existing portfolio materially diversifies or amplifies its risks.
Official syllabus reference: [1]
42. Control testing distinguishes design from operation
Control design asks whether a procedure, if performed as intended, addresses the specified risk. Operating-effectiveness testing asks whether it actually occurred with the required quality and frequency. A signed approval can demonstrate a recorded action without demonstrating a substantive review. Application defines the control objective, evidence, population, exceptions and remediation, then examines whether failures indicate isolated mistakes or a systematic weakness. Control cost and delays should also be considered. Stronger control does not necessarily mean more approval layers; clearer information, automation or segregation may address the causal weakness more directly.
Worked example: Payment approvals are consistently signed, but approvers never compare beneficiary requests with independent records. Testing identifies an operating failure if verification was required, or a design failure if the procedure never specified that comparison in the first place.
Common mistake: Approval signatures alone do not establish control effectiveness when the evidence does not show that the intended verification was actually performed.
Official syllabus reference: [1]
43. Futures hedges remove symmetric price exposure
A futures position creates approximately linear exposure to changes in its underlying price. A short position can offset losses on a matching asset when prices fall, but also offsets gains when prices rise. Hedge sizing should reflect contract sensitivity, exposure size and any imperfect relationship between them. Futures typically involve margin and frequent settlement, so economic offset does not ensure matching cash flows. Application assesses effectiveness, rollover, basis risk and collateral liquidity together. A hedge ratio estimated from historical covariance is a statistical starting point whose stability must be tested under relevant stress conditions.
Worked example: An insurer holds a matching equity-index portfolio and shorts index futures. Falling index values generate futures gains that offset asset losses; rising values generate margin payments even though the portfolio gains, creating a separate cash-management requirement.
Common mistake: A futures hedge can stabilize economic value while increasing short-term cash demands because derivative settlements and underlying gains need not arrive together.
Official syllabus reference: [1]
44. Protective puts preserve asymmetric participation
A purchased put provides a payoff equal to the positive difference between strike and underlying value at exercise, subject to contract terms. Combined with a matching asset, it limits downside below the strike at the relevant maturity while retaining upside, with the premium reducing net returns. Protection depends on matching the underlying, quantity and horizon. Delta describes local price sensitivity but changes as markets move, so an option's protection differs from a fixed linear hedge. Application compares strike, maturity, premium, liquidity and counterparty arrangements against the actual loss profile being managed.
Worked example: An insurer holds an index portfolio worth USD 100 million and buys matching puts with a USD 90 million strike-equivalent value. At maturity, a portfolio value of USD 75 million is supplemented by a USD 15 million payoff before premium.
Common mistake: A put's maturity floor does not automatically guarantee interim cash availability or protection for a portfolio that moves differently from the option's underlying.
Official syllabus reference: [1]
45. Basis risk is imperfect hedge alignment
Basis risk is residual exposure caused by differences between the hedged item and the instrument used to offset it. Differences can involve underlying assets, reference rates, maturity, geography, currency or contractual definitions. Historical correlation can deteriorate precisely when protection is needed. Application identifies the economic drivers of mismatch, measures residual exposure across scenarios and checks whether the proposed hedge still fits appetite. A proxy hedge can be valuable despite basis risk, but effectiveness should be evaluated on the combined position rather than inferred from the hedge instrument's standalone performance.
Worked example: An insurer hedges a portfolio of regional property shares with a broad equity index. The index falls 8% while the property portfolio falls 18%, leaving substantial loss because sector-specific financing pressures are not captured by the broad hedge.
Common mistake: A high historical correlation does not eliminate basis risk when exposures differ structurally or their relationship changes during stressed market conditions.
Official syllabus reference: [1]
46. Quota-share reinsurance transfers a proportion
Quota-share reinsurance cedes an agreed proportion of covered premiums and losses, subject to the treaty's detailed terms. It can reduce retained underwriting exposure and support capacity, but also transfers part of favorable experience and introduces counterparty dependence. Commissions and expense allowances affect the net economics and should not be ignored. Application calculates gross and ceded cash flows consistently, then evaluates retained profit, capital effects and recovery timing. A proportional treaty does not have the same loss-shaping effect as an excess-of-loss layer, which responds only after a specified attachment is reached.
Worked example: Under a simplified 30% quota-share treaty, USD 20 million of covered claims produces USD 6 million of ceded claims and USD 14 million retained. Premium cession and treaty commissions must also be included before assessing the arrangement's profitability.
Common mistake: A quota-share percentage does not describe the insurer's net profit reduction because premiums, expenses, commissions and capital effects also change.
Official syllabus reference: [1]
47. Excess-of-loss layers have finite protection
For a simplified excess-of-loss layer with attachment A and payment limit M, recovery on covered loss L is min(max(L − A, 0), M). Retained loss equals gross loss minus recovery. Losses below attachment remain with the insurer, and losses beyond attachment plus limit create additional retention above the exhausted layer. Application must also distinguish per-risk, per-event and aggregate structures, since they respond to different groupings of losses. Reinstatements, exclusions, expenses and recovery timing can materially alter actual results and should be specified rather than assumed.
Worked example: A USD 15 million layer attaching at USD 5 million pays USD 15 million on a USD 24 million covered event. The insurer retains USD 9 million: USD 5 million below attachment and USD 4 million above exhaustion.
Common mistake: The attachment is not a cap on retained loss when the treaty has a finite payment limit and gross losses exceed the protected layer.
Official syllabus reference: [1]
48. Reinsurance recovery creates counterparty exposure
Reinsurance changes the source of risk as well as its amount. A cedant may reduce insured-loss exposure while acquiring recoverable credit risk, payment delay, dispute risk and concentration in reinsurers. Wrong-way risk arises when the reinsurer's ability to pay deteriorates in the same conditions that create large recoveries. Application evaluates net loss under both insured events and counterparty stress, considering collateral quality, enforceability, limits and operational collection. Contractual recovery is not equivalent to immediate cash, and accounting recognition of an asset does not prove that it will fund obligations when due.
Worked example: A catastrophe produces a large reinsurance recoverable, but the same event weakens the reinsurer. The cedant tests delayed payment and partial recovery, including whether collateral can be accessed in time to meet its own claim obligations.
Common mistake: Deducting the full contractual recovery from gross loss can understate residual exposure when default, disputes, collateral restrictions or collection delays are material.
Official syllabus reference: [1]
49. Duration matching protects economic surplus locally
Modified duration approximates the percentage value change for a small parallel yield movement: ΔV ≈ −DVΔy. Economic-surplus sensitivity depends on asset dollar duration minus liability dollar duration, not merely on matching the two duration numbers. Application measures both sides on consistent valuation bases and sizes assets or swaps to offset the difference. The approximation is local and can fail for large moves or changing cash flows. Key-rate sensitivities extend the analysis to curve-shape changes, while liquidity and collateral consequences remain separate from the targeted economic-value offset.
Worked example: Assets lose USD 40,000 and liabilities fall USD 59,500 when yields rise one basis point. Surplus gains USD 19,500, so a receive-fixed swap losing that amount for the same movement locally offsets the gap.
Common mistake: Equal asset and liability durations do not ensure surplus neutrality when their market values differ, because dollar sensitivities depend on both duration and value.
Official syllabus reference: [1]
50. Convexity and options destabilize duration hedges
Convexity captures curvature in the relationship between value and yield, extending the duration approximation with a second-order term. Embedded options can make cash flows change as rates move, causing duration and convexity to shift materially. Callable assets, surrender options and guarantees can therefore create hedges that appear matched initially but diverge under larger stresses. Application revalues both assets and liabilities across relevant paths and considers behavioral assumptions. A static duration match is a local approximation, while option-sensitive ALM may require dynamic adjustment and explicit recognition of transaction costs and liquidity constraints.
Worked example: An insurer matches initial asset and liability duration. When rates fall, callable bonds are redeemed and surrender behavior changes, altering both cash-flow profiles. Full scenario revaluation reveals a surplus loss that the initial duration calculation did not predict.
Common mistake: An initial duration match does not lock in protection when embedded options or customer behavior change the timing and amount of future cash flows.
Official syllabus reference: [1]
51. Operational resilience targets service continuity
Operational resilience focuses on maintaining or restoring important services through disruption. It begins with service outcomes, then maps people, processes, technology, facilities and external dependencies needed to deliver them. Recovery objectives should be internally justified by consequences and tested against realistic disruption, rather than treated as universal thresholds. Application tests end-to-end service restoration, data integrity and manual workarounds, including resource conflicts during simultaneous failures. Insurance can fund some losses but cannot itself restore a service. Resilience decisions therefore compare redundancy, substitutability and recovery capability against customer and enterprise consequences.
Worked example: An insurer tests its ability to pay urgent claims while its main administrator is unavailable. The exercise includes alternate authorization, bank access and accurate customer records, revealing that a restored website alone would not restore the payment service.
Common mistake: Recovering an individual application does not establish service resilience when authorization, data, staffing or external payment dependencies remain unavailable.
Official syllabus reference: [1]
52. Adaptive strategies use signposts and options
Under deep uncertainty, a robust strategy seeks acceptable outcomes across several plausible futures rather than optimizing only one forecast. Adaptive planning identifies actions available now, options preserved for later and signposts that trigger a change in course. This is especially useful for climate, technology and strategic risks where probabilities and timing are uncertain. Application compares irreversible commitments with staged decisions and specifies observable triggers before uncertainty resolves. Flexibility has a cost, so preserving options should be evaluated against foregone returns and the likelihood that delayed action would become ineffective or prohibitively expensive.
Worked example: Instead of immediately replacing an entire coastal property portfolio, an insurer stages new investment restrictions and watches flood-insurance availability, occupancy and policy developments. Predetermined signposts trigger stronger action while diversification preserves flexibility during the uncertain transition over time.
Common mistake: Adaptive planning is not indefinite postponement; it requires explicit signposts, prepared actions and recognition that some opportunities to respond expire.
Official syllabus reference: [1]
53. Economic capital models begin with purpose
An economic capital model estimates resources needed to support risks under an internally specified risk measure, horizon and valuation basis. Design begins with intended decisions and material exposures, then specifies data, distributions, dependencies, management actions and aggregation. Available capital must be defined consistently with the requirement. Application balances sophistication against transparency and decision relevance, validates key sensitivities and communicates uncertainty. A one-year revaluation model and a multi-year runoff model answer different questions; their outputs should not be compared as though differing results necessarily reveal an error.
Worked example: An insurer designs a one-year surplus-loss model for planning and allocation. It includes market, insurance and operational exposures, revalues remaining obligations at the horizon and separately tests liquidity, which the capital statistic alone does not measure.
Common mistake: Selecting a confidence level before defining the model's purpose, horizon and valuation basis can produce precise-looking capital that answers the wrong question.
Official syllabus reference: [1]
54. Capital and value measures serve different users
Accounting equity follows reporting recognition and measurement rules. Regulatory capital resources and requirements follow the applicable supervisory regime. Economic capital reflects an internal assessment of risk, while economic value measures the value of resources and obligations under specified assumptions. Rating-agency perspectives can add another constraint. Differences can arise from valuation, eligible resources, stresses, diversification and horizon, so one figure should not be substituted for another without reconciliation. Application identifies the binding constraints for a decision and explains the bridges between measures, avoiding unsupported assumptions that a higher value under one basis implies stronger protection under every basis.
Worked example: A transaction improves reported earnings but increases internal tail risk and required regulatory resources. Management reconciles these effects before approving it, recognizing that an accounting gain does not automatically provide usable capital or improve the enterprise's risk-adjusted economic position.
Common mistake: Accounting profit, economic value creation and regulatory capital release are different outcomes and need separate calculations under their relevant measurement bases.
Official syllabus reference: [1]
55. Capital definitions must specify expected losses
A tail loss measure can be used as total required resources or as a basis for unexpected-loss capital, depending on the definition. If expected losses are already funded through pricing or provisions, a model may define unexpected-loss capital as a selected risk measure minus expected loss. Other frameworks use different constructions. Application states the convention, avoids double counting and aligns available resources with the modeled loss variable. VaR and TVaR can produce materially different requirements for heavy tails, so the choice should reflect appetite and purpose rather than convenience alone.
Worked example: A simplified model has expected annual loss of USD 3 million and 99% TVaR of USD 20 million. If its explicit convention funds expected loss separately, unexpected-loss capital is USD 17 million; total tail resources remain USD 20 million.
Common mistake: Subtracting expected loss is not a universal capital rule; it is appropriate only when the defined framework funds that component separately and consistently.
Official syllabus reference: [1]
56. Embedded value separates existing business components
Embedded value measures shareholder value associated with covered existing business using a specified methodology. A common decomposition combines adjusted net worth with the present value of future distributable profits, allowing for relevant costs such as required capital and options or guarantees. Methods differ, so components and discounting must be identified before comparing results. It is not automatically a market sale price or the value of future uncontracted new business. Application tests sensitivities to assumptions and avoids counting required capital or its release twice when moving between net worth and future-profit components.
Worked example: Under a stated simplified method, adjusted net worth is USD 40 million, future-profit value is USD 30 million and additional option and capital costs are USD 8 million. Embedded value is USD 62 million for the covered existing business.
Common mistake: Embedded value does not automatically include all future new business or equal the price a buyer would pay for the entire company.
Official syllabus reference: [1]
57. RAROC and EVA compare return with capital cost
RAROC divides a consistently defined risk-adjusted profit measure by allocated capital. EVA subtracts a capital charge from profit, commonly expressed as profit minus the required return times capital. RAROC measures a rate, while EVA measures value in currency over the stated period. Application aligns tax treatment, timing, expected-loss adjustments and capital definitions, ensuring that the numerator has not already deducted the same charge. A high ratio on a small activity can create less total value than a lower ratio on a larger activity, especially when scaling opportunities differ.
Worked example: Annual profit before capital cost is USD 9 million, allocated capital is USD 60 million and the required return is 12%. RAROC is 15%, the capital charge is USD 7.2 million and EVA is USD 1.8 million.
Common mistake: The difference between RAROC and the hurdle rate is a return spread, not EVA until it is multiplied by the relevant capital amount.
Official syllabus reference: [1]
58. Euler allocation uses marginal sensitivities
For a differentiable capital function homogeneous of degree one, Euler allocation assigns exposure i the amount xi multiplied by the partial derivative of total capital with respect to xi. Euler's theorem makes these contributions sum to total capital under those assumptions. The method links allocation to local aggregate-risk sensitivity rather than distributing capital solely by standalone size. Application checks differentiability, scaling definitions and numerical stability. An allocated amount need not equal the capital saved by completely removing a unit, because a local derivative and a finite portfolio change are different calculations.
Worked example: For K = √(x² + y² + 2ρxy), x = 6, y = 8 and ρ = 0.25, x's allocation is 6(6 + 2)/√124, approximately USD 4.31 million when inputs are in USD millions.
Common mistake: Euler capital is not generally the same as standalone capital, proportional allocation or the finite reduction obtained by removing the whole business unit.
Official syllabus reference: [1]
59. Marginal, incremental and standalone capital differ
Standalone capital measures a unit in isolation. Incremental capital measures the finite change in enterprise capital when a unit or project is added or removed. Marginal capital measures local sensitivity to a small change in exposure, with terminology requiring explicit definition. These quantities differ because diversification and concentration depend on the rest of the portfolio. Application selects the measure suited to pricing, project approval, allocation or exit analysis. Do not assume removal benefits equal entry costs in a changed portfolio, or that finite incremental contributions across all units necessarily sum to total capital.
Worked example: A unit has USD 10 million standalone capital. Enterprise capital falls from USD 50 million to USD 44 million when it is removed, giving a USD 6 million removal increment; a local scaling derivative may produce another value.
Common mistake: Using the word marginal without defining whether it means a derivative or a finite portfolio change can obscure materially different capital calculations.
Official syllabus reference: [1]
60. Risk-cost attribution avoids double charging
Performance attribution should identify the costs of expected losses, mitigation, funding and capital without counting the same economic burden twice. Reinsurance premiums, hedge costs and central liquidity resources may benefit several units, requiring a transparent allocation driver. A common enterprise capital rate supports comparability, while a unit-specific rate needs a clear rationale for risks or financing costs not already captured elsewhere. Application reconciles allocated charges to total enterprise costs and tests incentives, including whether units can improve reported performance by shifting exposure to an uncharged central function rather than reducing enterprise risk.
Worked example: A central hedge protects two units. The insurer allocates its premium by measured protection benefit and separately charges residual capital. It checks that the profit numerator has not already deducted the same premium through another internal transfer price.
Common mistake: A higher capital charge does not improve attribution when it duplicates risks already reflected in reduced profit, mitigation premiums or the capital amount itself.
Official syllabus reference: [1]