You’re on a roll
Create a free account to save your FRM progress, track your accuracy by topic, and pick up right where you left off.
Free Questions › FRM › Operational Risk and Resilience
Free FRM Operational Risk and Resilience Practice Questions & Answers
400 exam-style Operational Risk and Resilience questions. Pick your answer, hit Check answer, and see the worked solution — free to start, no signup.
100% free · No login to startQuestion 31
What is 'recovery' in operational loss data and why is it important?
Select an option first.
Correct answer: B — The amount of an operational loss that is subsequently recovered — through insurance payouts, legal recoveries, or the reversal of erroneous transactions. The 'net loss' = gross loss − recovery. Accurate tracking of recoveries is important for regulatory reporting and for calibrating the expected net loss in capital models
Explanation: B is correct. Recovery in OpRisk: gross loss is the full initial impact; net loss is after recoveries. Insurance recoveries: most banks carry insurance for certain OpRisk events (crime insurance, professional indemnity). Under AMA, regulators allowed recognition of insurance as a capital mitigant (up to 20% capital reduction). Under Basel IV SA, insurance is not explicitly recognised. Recovery tracking improves loss data quality and avoids overstating actual net losses. A is recovery time (BCP concept). C and D are incorrect.
Question 32
What is a 'rogue trader' event and what operational risk categories does it typically involve?
Select an option first.
Correct answer: B — A trader who deliberately conceals unauthorised trading positions that exceed their limits — typically involving: internal fraud (unauthorised activity, falsification of records) and execution/process management failures (failures of position monitoring and reconciliation). Famous examples: Nick Leeson (Barings 1995), Jérôme Kerviel (SocGen 2008), Bruno Iksil 'London Whale' (JPMorgan 2012)
Explanation: B is correct. Rogue trader: both a people failure (deliberate misconduct) and a process/control failure (inadequate position monitoring, failure of segregation of duties, inadequate oversight). The Basel event categories involved: Internal Fraud (unauthorised activity) and Execution/Delivery/Process Management (reconciliation failures, limit monitoring failures). A is an execution quality issue. C is external fraud. D is model risk.
Question 33
What is 'model risk' as a component of operational risk?
Select an option first.
Correct answer: B — The risk of material loss or adverse outcome from decisions based on incorrect, misused, or inappropriately applied quantitative models — includes: model errors (incorrect mathematical specification), calibration errors (wrong input data), implementation errors (coding bugs), and misuse (applying a model outside its intended range or without proper understanding
Explanation: B is correct. Model risk (OCC/Fed SR 11-7, EBA): classified under Execution/Delivery/Process Management or Clients/Products/Business Practices in the Basel framework, depending on the consequences. Sources: (1) Conceptual errors in the model design; (2) Data errors; (3) Coding/implementation bugs; (4) Inappropriate use of the model. Famous model failures: CDO Gaussian copula (2008), Knight Capital algorithm ($440M loss in 45 minutes, 2012). A is random error. C is incorrect. D: model risk affects all areas (lending, pricing, OpRisk capital itself).
Question 34
What is the 'tail of the operational loss distribution' and why is it so difficult to model accurately?
Select an option first.
Correct answer: B — The extreme right tail of the aggregate annual loss distribution — representing rare, high-impact events like large rogue trading losses, major cyber incidents, or regulatory fines exceeding $1B. Difficult to model because: (1) Very few historical data points; (2) High parameter uncertainty in heavy-tailed distributions; (3) The tail is driven by unique, one-off events that may not recur; (4) External data scaling introduces additional uncertainty
Explanation: B is correct. Tail modelling challenges: the OpRisk capital requirement is the 99.9th percentile (1-in-1000-year event). With only 5-10 years of internal data, you might have zero or one observation in the tail region. This means the capital estimate is highly sensitive to distributional assumptions. Small changes in the Pareto shape parameter (ξ) can change the 99.9th percentile capital by 50%+. A is the body (high-frequency/low-severity). C and D are other distribution features.
Question 35
What is 'insurance as an operational risk mitigant' and how is it recognised under different capital frameworks?
Select an option first.
Correct answer: B — Under Basel II AMA, banks could receive up to 20% reduction in operational risk capital through recognised insurance policies (e.g., bankers blanket bond, professional indemnity, D&O, property insurance) — subject to: minimum credit rating of insurer (A-), policy terms meeting specific requirements, and regulatory approval. Under Basel IV SA, explicit insurance capital recognition is not provided
Explanation: B is correct. Insurance and OpRisk capital: AMA allowed 20% max capital reduction for qualifying insurance. Conditions: insurer rated A- or better, policy covers the loss type, remaining term > 1 year, no exclusions undermining coverage. Basel IV SA: dropped formal insurance recognition but banks still use insurance to manage net losses. Insurance remains a risk management tool regardless of capital treatment. A and C are wrong. D is incorrect — insurance is a key OpRisk management technique.
Get the full FRM question bank — free
Drop your email and we'll send you fresh FRM practice questions, fully worked solutions and exam-deadline reminders. No spam — unsubscribe in one click.
Want to save your score and take a full mock exam? Create a free account →
Question 36
What is 'stress testing' in operational risk and how does it differ from scenario analysis?
Select an option first.
Correct answer: B — Stress testing evaluates the impact of extreme adverse events on a bank's capital position and financial health — typically regulatory-driven (e.g., DFAST, EBA stress tests). Scenario analysis in OpRisk is typically more granular and process-focused (identifying specific operational loss scenarios for capital modelling), while stress testing is portfolio-level and macro-focused
Explanation: B is correct. Distinction: scenario analysis (OpRisk): forward-looking, expert-led, focuses on specific operational events, feeds into the capital model directly. Stress testing: top-down, regulator-driven, assesses overall capital adequacy under macro-stress scenarios. OpRisk stress testing: what is the additional operational loss under a severe recession scenario? (Operational losses tend to rise in economic downturns — more fraud, more litigation.) A conflates them. C is not a defining distinction. D: stress testing applies to all risk types.
Question 37
What is the 'Data Collection Exercise' (DCE) conducted by the Basel Committee?
Select an option first.
Correct answer: B — A periodic industry-wide data collection exercise where large banks submit anonymised operational loss data to the Basel Committee, allowing calibration of the Basel SA capital multipliers and benchmarking of banks' loss profiles across different business lines and event types
Explanation: B is correct. BCBS DCE: the Basel Committee periodically collects loss data from participating banks to: understand the distribution of operational losses globally, calibrate the BI multiplier rates (α) in the Basel IV SA, and assess the appropriateness of the ILM. Data is anonymised and aggregated. The DCE has informed multiple rounds of Basel capital calibration. A is a data governance survey. C is a supervisory inspection. D: data is anonymised, not public.
Question 38
What is 'economic capital' (EC) for operational risk and how does it differ from regulatory capital?
Select an option first.
Correct answer: B — Economic capital is the bank's own internal estimate of the capital needed to cover unexpected operational losses at a high confidence level (e.g., 99.97% to match AA credit quality) using the bank's own risk assessment methods. Regulatory capital (Basel SA) uses standardised formulae. EC may be higher or lower than regulatory capital depending on the bank's risk profile and modelling assumptions
Explanation: B is correct. EC vs regulatory capital for OpRisk: Economic capital uses the bank's own models (often LDA-based) at a confidence level matching the target credit rating (99.97% for AA). Regulatory capital uses the standardised approach. Pre-Basel IV, AMA banks often had lower regulatory capital than their EC estimates. Post-Basel IV SA, some banks find regulatory capital exceeds economic capital (SA is deliberately conservative). A: direction varies. C: they use different methods. D: EC applies to all risk types.
Question 39
What is 'risk aggregation' in operational risk and what are the key challenges?
Select an option first.
Correct answer: B — Combining operational risk capital estimates across different business lines, risk categories, and legal entities to obtain a firm-wide OpRisk capital figure — challenges include: choosing appropriate correlation assumptions (independence vs perfect correlation), accounting for diversification benefits, aggregating across different currencies, and combining qualitative (RCSA) and quantitative (LDA) risk assessments
Explanation: B is correct. Aggregation challenges: (1) Correlation assumptions — how correlated are losses across different event types and business lines? Most banks assume some diversification benefit (correlations 0-0.75 rather than 1). (2) Diversification vs concentration — a bank concentrated in one business line has less diversification. (3) Qualitative vs quantitative integration — RCSA scores inform capital adjustments but the linkage is subjective. (4) Top-down vs bottom-up consistency. A is KRI reporting. C and D are oversimplified approaches.
Question 40
What is 'risk transfer' in the context of operational risk management?
Select an option first.
Correct answer: B — Shifting the financial consequence of operational risk events to third parties through: insurance (transferring specific event risk for a premium), outsourcing contracts (transferring operational responsibility), indemnification clauses, and hedging. However, risk transfer often does not eliminate reputational risk or regulatory responsibility
Explanation: B is correct. OpRisk transfer: insurance is the most common tool. Banks buy: Bankers Blanket Bond (internal fraud, external fraud), Professional Indemnity (negligence, errors), D&O (directors and officers liability), Cyber Insurance (cyber incidents). Outsourcing: operations transferred to a third party shift execution risk but not regulatory/reputational risk. Capital markets risk transfer for OpRisk is limited (catastrophe bonds are mainly for insurance companies). A is organisational. C is an internal management decision. D: capital markets OpRisk transfer is nascent.
Question 41
What is the 'compound Poisson process' and why is it used in LDA modelling?
Select an option first.
Correct answer: B — A stochastic process where the number of loss events follows a Poisson distribution and each event has a random loss amount drawn from a severity distribution — by compounding the frequency and severity, we obtain the aggregate annual loss distribution. This is the mathematical foundation of the LDA in operational risk capital modelling
Explanation: B is correct. Compound Poisson: S = X₁ + X₂ + ... + X_N, where N ~ Poisson(λ) and Xᵢ ~ Severity distribution. The characteristic function of S can be computed analytically, but the full distribution is typically derived numerically via Monte Carlo or Panjer recursion. Key property: the mean of S = λ × E[X]; variance of S = λ × E[X²]. A is not the compound Poisson definition. C is just scaling. D: used for all OpRisk categories.
Question 42
What is the 'Panjer recursion' method in LDA modelling?
Select an option first.
Correct answer: B — A numerical algorithm that efficiently computes the aggregate loss distribution by exploiting the recursive structure of the Poisson (or Binomial/Negative Binomial) frequency distribution combined with a discretised severity distribution — computationally faster than Monte Carlo for many risk cells
Explanation: B is correct. Panjer recursion: P(S = k) = Σ (a + b·j/k) × P(S = k-j) × f(j), where f(j) is the discretised severity probability. This recursion builds up the aggregate distribution from the convolution of frequency and severity. Advantages: computationally efficient, no sampling error (unlike Monte Carlo). Limitations: requires discretisation of the severity distribution and can be slow for very heavy-tailed distributions. A is distribution fitting. C is the expectation formula (simple calculation). D is reading off the distribution (not a method).
Question 43
What is 'data quality' in operational risk modelling and why is it critical?
Select an option first.
Correct answer: B — The accuracy, completeness, consistency, timeliness, and relevance of the data used in risk models and management information — critical because: 'garbage in, garbage out' — poor quality data produces unreliable capital estimates, misleading KRI dashboards, and incomplete RCSA assessments. BCBS 239 requires G-SIBs to have robust risk data aggregation capabilities
Explanation: B is correct. Data quality dimensions: Accuracy (data is correct — losses are properly classified, amounts are accurate). Completeness (all qualifying events are captured — no gaps). Consistency (same definitions applied across business units and time). Timeliness (data is available when needed). Relevance (data reflects current risk profile, not just historical conditions). Poor data quality in the loss database produces biased LDA parameters and understated capital. A is aesthetics. C and D are too narrow.
Question 44
What is 'mixture distribution' in operational risk severity modelling?
Select an option first.
Correct answer: B — A distribution formed by combining two or more component distributions with different weights — used for operational risk severity to capture both the body (frequent, smaller losses modelled with lognormal) and the tail (rare, large losses modelled with Pareto/GPD). The mixture provides better fit across the full range of observed losses
Explanation: B is correct. Mixture severity distribution: f(x) = p × f₁(x) + (1-p) × f₂(x), where f₁ = body distribution (lognormal), f₂ = tail distribution (Pareto), p = mixing weight. This 'spliced distribution' handles both the empirical body and the extreme tail more accurately than a single distribution. Parameters estimated jointly via MLE. Advantage: body fit doesn't compromise tail fit. A mixes different distribution types. C is a data combination. D is external data scaling.
Question 45
What is the 'lognormal distribution' and why is it commonly used for operational risk severity?
Select an option first.
Correct answer: B — A probability distribution where the logarithm of the variable follows a normal distribution — characterised by two parameters: μ (log mean) and σ (log standard deviation). Used for OpRisk severity because: it is positively skewed (most losses small, few large), all values are positive (losses can't be negative), and it is mathematically tractable. However, its tail is lighter than many empirical OpRisk loss distributions
Explanation: B is correct. Lognormal severity: if X ~ Lognormal(μ, σ²), then ln(X) ~ Normal(μ, σ²). Mean = e^(μ + σ²/2), Variance = (e^σ² - 1) × e^(2μ + σ²). Common choice because it fits the body of OpRisk data well. Limitation: the tail of the lognormal is lighter than empirical OpRisk data in many cases — Pareto or GPD provide better tail fits. A describes left-skewed. C is frequency distribution (Poisson). D: lognormal has a heavier tail than normal.
Question 46
What is the 'Pareto distribution' and what is its key property for tail modelling?
Select an option first.
Correct answer: B — A heavy-tailed distribution with a power law tail — P(X > x) ~ (x/x_min)^(-α) for large x. Key property: the tail decays as a power law rather than exponentially, meaning extreme events are much more likely than a normal or lognormal distribution would predict. The Pareto distribution is a special case of the Generalised Pareto Distribution with shape parameter ξ > 0
Explanation: B is correct. Pareto distribution: scale-free (self-similar) — a Pareto-distributed loss that is twice as large than the threshold is also Pareto distributed. Example: the top 20% of causes produce 80% of effects (Pareto principle, 80/20 rule). For OpRisk: the largest losses can be orders of magnitude larger than typical losses. Pareto index α: lower α = heavier tail = more extreme losses relatively more likely. A: Pareto has no upper bound. C: Pareto is used for all tail-heavy OpRisk categories. D: Pareto has a heavier tail than lognormal.
Question 47
What is 'value at risk' at the 99.9th percentile in operational risk and what does it represent?
Select an option first.
Correct answer: B — The loss level that will only be exceeded with probability 0.1% over a one-year horizon — meaning in any given year, there is a 99.9% probability that operational losses will be below this level and a 0.1% probability they will exceed it. Under Basel, this is the standard for operational risk regulatory capital
Explanation: B is correct. 99.9th percentile VaR: corresponds to a 1-in-1000-year event. Basel selected this to align with the 99.9% confidence level used for credit risk, reflecting the target of a bank being able to survive a 1-in-1000-year operational loss. The 99.9th percentile is in the extreme tail — estimated via LDA Monte Carlo, EVT, or scenario analysis. A is expected loss (mean). C is expected loss. D reverses the probability — 99.9% probability losses will be BELOW this level.
Question 48
What is the 'actuarial method' in operational risk and how does it relate to LDA?
Select an option first.
Correct answer: B — The general actuarial approach of modelling aggregate claims as the sum of a random number of individual claim amounts — mathematically identical to the LDA in operational risk. Actuarial science provides the mathematical foundation for the compound frequency-severity model, including: moment-based methods, Panjer recursion, and the theory of extreme values. The insurance industry has applied these techniques for over a century
Explanation: B is correct. Actuarial OpRisk: the LDA is borrowed directly from actuarial casualty insurance modelling (collective risk model). The frequency-severity compound model, extreme value theory, and credibility weighting are all actuarial techniques applied to OpRisk capital modelling. This connection provides a rich theoretical and practical literature from insurance that OpRisk practitioners draw on. A: the actuarial method is used by both insurers and banks. C: insurance pricing is a related but separate application. D: the full LDA includes the full distribution, not just moments.
Question 49
What is 'extreme value theory' (EVT) and what is its main result for tail modelling?
Select an option first.
Correct answer: B — A branch of probability theory that studies the behaviour of maxima and extreme values of random variables — the key result (Extremal Types Theorem, Fisher-Tippett-Gnedenko): the maximum of a large sample of iid random variables converges to one of three extreme value distributions (Gumbel, Fréchet, Weibull) regardless of the original distribution. For sums (aggregate losses), the Generalised Pareto Distribution (GPD) applies to exceedances above a threshold
Explanation: B is correct. EVT for OpRisk: the key practical result is that for a wide class of underlying distributions, the distribution of exceedances above a high threshold converges to the GPD (Pickands-Balkema-de Haan theorem). This provides theoretical justification for using GPD to model the tail of operational loss data — it is asymptotically correct. A describes event frequency. C: EVT provides a statistical framework, not historical analysis. D: different underlying distributions lead to different tail behaviour.
Question 50
What is the 'negative binomial distribution' and when is it used instead of Poisson for frequency modelling?
Select an option first.
Correct answer: B — A frequency distribution used when operational loss counts show over-dispersion — variance greater than mean. The Negative Binomial has two parameters (r, p) and variance = mean + mean²/r. When the Poisson assumption (variance = mean) is violated, NB provides a better fit. Over-dispersion commonly arises when loss frequency itself varies over time (e.g., fraud rates spike during economic downturns)
Explanation: B is correct. Negative Binomial frequency: if the mean of a Poisson distribution varies (λ itself is random, e.g., gamma-distributed), the marginal distribution of counts follows a Negative Binomial. This is common in OpRisk — fraud rates are not constant but vary with economic conditions and control quality. Testing for over-dispersion: compare sample variance to sample mean; if variance >> mean, Poisson underestimates frequency variability → NB is preferred. A: losses themselves are non-negative; the distribution describes counts. C: Poisson is still commonly used for simplicity. D: NB models counts, not loss sizes.
Question 51
What is 'risk transfer through securitisation' in the context of operational risk?
Select an option first.
Correct answer: B — A theoretical approach where operational risk exposures are packaged into securities that transfer the risk to capital market investors — analogous to credit risk securitisation (CLOs). In practice: operational risk securitisation is extremely limited because: operational losses are harder to define and document than credit defaults, moral hazard (insurance removes incentive to manage risk), basis risk (the investor's payout may not match actual losses), and market appetite is thin
Explanation: B is correct. OpRisk securitisation: unlike credit risk (CLOs have a large established market), operational risk securitisation barely exists in practice. Cat bonds (catastrophe bonds for insurance companies) are the closest analogue — they transfer specific catastrophe risk to bond investors. The absence of a liquid OpRisk securitisation market means banks rely on insurance, self-insurance (capital), and risk mitigation rather than risk transfer through capital markets. A is an accounting treatment. C is insurance. D is a regulatory capital reduction method.
Question 52
What is 'self-insurance' in operational risk management?
Select an option first.
Correct answer: B — Retaining operational risk within the organisation rather than transferring it to an insurer — the bank effectively 'insures itself' by holding capital and building reserves. Self-insurance is rational when: insurance premiums exceed the actuarially fair premium by too much, the bank has a large diversified portfolio of OpRisk (diversification benefit), or insurance is unavailable for specific risk types
Explanation: B is correct. Self-insurance logic: for large, well-diversified banks, the insurance premium includes: actuarially fair premium + insurer profit margin + adverse selection premium (insurer thinks the bank knows more about its risks). If the bank can hold capital at a lower cost than paying insurance premiums, self-insurance is economically rational. Capital holding: economic capital absorbs unexpected losses (self-insurance). Expected losses: built into pricing and provisions. A is internal investigation. C is zero-insurance (may be self-insurance but imprecisely described). D is internal transfer pricing.
Question 53
What is the 'principle of proportionality' in Basel operational risk regulation?
Select an option first.
Correct answer: B — The Basel framework should be applied in a manner proportionate to the size, complexity, and risk profile of the bank — smaller, less complex banks use simpler approaches (BIA/SA) with less granular risk identification. Larger, more complex banks (G-SIBs) are expected to have sophisticated ORMF, granular data, and advanced capital modelling
Explanation: B is correct. Proportionality: the Basel IV SA applies to all banks but is simpler (no internal models). However, the quality and sophistication of the operational risk management framework is expected to scale with the bank's size and complexity: G-SIB: advanced RCSA, comprehensive KRI framework, detailed LDA model, board-level OpRisk reporting. Small bank: basic loss database, simple RCSA, limited KRIs. Supervisors apply proportionality in their assessment intensity. A: sophisticated approaches are for large banks, not all. C: not strictly true (risk profiles matter too). D is the BIA formula description.
Question 54
What is the 'data windowing' problem in operational risk modelling?
Select an option first.
Correct answer: B — The challenge of selecting the appropriate historical time period for calibrating the LDA model — a longer window (10 years) provides more data points but may include events no longer representative of current risk (major system changes, business model changes). A shorter window (3-5 years) is more representative but provides insufficient data for tail estimation
Explanation: B is correct. Data window selection: tension between relevance (recent data reflects current risk) and sufficiency (tail estimation needs many data points). Solutions: (1) Weighted data (recent events weighted more heavily); (2) Regime analysis (identify and separate different historical risk regimes); (3) Combination with scenario analysis (use historical data for frequency/body, scenarios for tail). Basel IV SA: ILM uses 10 years of loss data for large banks. A is IT display. C is a database design issue. D is a regulatory requirement for one specific purpose.
Question 55
What is 'the internal loss multiplier' (ILM) and what does a high ILM indicate about a bank?
Select an option first.
Correct answer: B — A component of the Basel IV SA capital formula: ILM = ln(exp(1) − 1 + (LC/BIC)^0.8), where LC = Loss Component (average annual 10-year losses × 15) and BIC = Business Indicator Component. An ILM > 1 indicates the bank has experienced more losses than the industry average for its size — capital is increased. ILM < 1 indicates fewer losses than average — for most banks, ILM is set to a minimum of 1 (floor applies)
Explanation: B is correct. ILM interpretation: ILM > 1 = above-average loss experience relative to BI → higher capital. ILM = 1 = at or below average (the floor prevents capital reduction for banks with good loss histories). This design addresses the concern that AMA allowed banks with manipulated models to reduce capital — the ILM ties capital to actual losses, not just model outputs. High ILM: the bank should investigate causes of elevated losses and implement control improvements. A reverses the relationship. C: ILM doesn't use market performance. D: applied to the entire book.
Question 56
What is 'tail dependence' in the context of operational risk aggregation?
Select an option first.
Correct answer: B — The tendency for extreme losses in one risk cell to occur simultaneously with extreme losses in other cells — standard correlation measures may underestimate this joint tail behaviour. Copulas with tail dependence (Clayton for lower tail, Gumbel for upper tail) can model this. High tail dependence means diversification benefits disappear precisely when they are most needed (during major loss events)
Explanation: B is correct. Tail dependence in OpRisk: a major cyber attack might simultaneously cause losses in: systems failure (downtime), external fraud (exploitation during outage), and clients/products (failed trades, customer harm). These losses are tail-dependent — they all spike together in the extreme event. Standard Gaussian copula assumes no tail dependence (correlation only in the body). Clayton/Gumbel copulas capture asymmetric tail dependence. Capital under tail dependence is higher than under Gaussian assumptions. A is left-tail concentration. C is frequency-severity dependency. D is historical frequency.
Question 57
What is the '99.9% VaR' capital confidence level in operational risk and how was it chosen?
Select an option first.
Correct answer: B — The 99.9% confidence level was selected for operational risk capital to be consistent with the credit risk capital confidence level (also 99.9%) and to reflect the level of protection needed to maintain a bank as a going concern through a 1-in-1000-year loss event. This corresponds to approximately the credit quality of an AA-rated institution
Explanation: B is correct. 99.9% rationale: Basel selected 99.9% for credit risk to target a AA-rated credit profile for banks (1-in-1000 insolvency probability per year). The same confidence level was applied to OpRisk and market risk for consistency and simplicity. Critics argue that OpRisk distributions are so uncertain at the 99.9th percentile that the confidence level is illusory — the estimate is highly imprecise. A: it was a policy decision with rationale. C and D are incorrect.
Question 58
What is the 'marginal contribution to capital' of a risk cell in an operational risk model?
Select an option first.
Correct answer: B — The change in total portfolio OpRisk capital resulting from a marginal increase in the risk of a specific cell — used to understand which risk cells are consuming the most capital and to guide risk management priorities. Cells with high marginal contribution should receive the most control investment if the goal is to reduce capital efficiently
Explanation: B is correct. Marginal capital contribution: analogous to marginal VaR in market risk. For correlated risks: the marginal contribution of a high-frequency/high-severity cell with high tail dependence to other cells is greater than the standalone capital would suggest. This information guides: investment in controls (target highest marginal contributors), business decisions (exit high marginal contribution activities if economic return is insufficient), and capital allocation to business units. A is minimum cell size. C is standalone capital. D is regulatory capital.
Question 59
What is 'forward-looking risk assessment' in operational risk and why is it complementary to historical data analysis?
Select an option first.
Correct answer: B — An assessment of emerging and future operational risks using forward-looking information — including: planned business changes (new products, new geographies, new technology), regulatory environment changes, industry threat intelligence, and scenario analysis for novel risk types. Complements historical data because past losses don't capture risks from new activities or emerging threats that haven't yet materialised
Explanation: B is correct. Forward-looking OpRisk: the RCSA explicitly incorporates forward-looking elements — when a business unit plans a major IT migration, the RCSA should flag the elevated technology risk before the migration happens. Scenario analysis: models future events that haven't occurred yet. BEICAFs (AMA) explicitly allowed forward-looking adjustments. New product approvals: the OpRisk function should assess OpRisk implications before product launch, not after losses have occurred. A is prediction (impossible). C: historical data remains essential. D is inflation adjustment.
Question 60
What is 'the expectation of the aggregate loss distribution' and how is it calculated from frequency and severity?
Select an option first.
Correct answer: B — For a compound Poisson model: E[S] = E[N] × E[X] = λ × μ_severity. The expected total loss per year equals the expected number of events (λ) multiplied by the expected individual loss amount (μ_severity). This is relatively easy to calculate and represents the 'expected loss' (EL) for the risk cell
Explanation: B is correct. Expected aggregate loss: E[S] = E[N] × E[X] for any compound distribution (not just Poisson). This is Wald's identity. Example: if expected frequency λ = 10 events/year and expected severity μ = $50,000 per event, then E[S] = 10 × $50,000 = $500,000 per year. This becomes the expected loss (EL) that should be priced into business line charges. A: the mean has a simple closed-form solution. C is incorrect — they are multiplied. D: Monte Carlo is needed for the full distribution, not just the mean.
More free FRM topics
Ten questions in
- The ones you miss are saved as a drill you can repeat
- Your place is kept, on this device and any other
- A streak, if that is the thing that gets you back tomorrow
Every question on this page stays free and open either way.