I85PREP
⌘ K
AIDA study materials
AIDA study resources

AIDA practice questions

One question at a time. Build your AIDA understanding.

First 50 of 200 questions. Each one lists the choices, the correct answer, and an explanation. The checker at the top of this page saves progress in this browser when it is available.

Continue in the practice tool above to load more questions, 50 at a time.

Question 1Using Data Analytics to Strengthen the Insurance Value Chain

Which stage of the insurance value chain produces the loss experience data most directly used to build predictive underwriting and pricing models?

  1. Investment management
  2. Product development
  3. Marketing and distribution
  4. Claims management Correct answer.

Explanation

Claims management records the timing, amount, and circumstances of losses, which become the outcome variable for frequency, severity, and risk-scoring models. Marketing, product design, and asset management generate other information but do not produce incurred loss outcomes that such models learn from.

Question 2Using Data Analytics to Strengthen the Insurance Value Chain

A report that summarizes average claim severity by region and coverage line for each of the last twelve months is an example of which type of analytics?

  1. Prescriptive analytics
  2. Predictive analytics
  3. Diagnostic analytics
  4. Descriptive analytics Correct answer.

Explanation

Descriptive analytics condenses historical data into summaries and trends, answering what happened. Diagnostic analytics goes further to explain why a pattern occurred, while predictive and prescriptive analytics project future outcomes and recommend actions.

Question 3Using Data Analytics to Strengthen the Insurance Value Chain

In a policy administration system, a large share of records has a blank occupation field. Which data quality dimension is most directly compromised?

  1. Accuracy
  2. Timeliness
  3. Completeness Correct answer.
  4. Uniqueness

Explanation

Completeness measures the proportion of required values that are actually populated, so a blank field is a completeness defect. The value is missing rather than wrong, late, or duplicated.

Question 4Using Data Analytics to Strengthen the Insurance Value Chain

An analytics tool evaluates every open bodily injury claim and outputs a recommended settlement range that minimizes expected total claim cost. This capability is best classified as:

  1. Diagnostic analytics
  2. Prescriptive analytics Correct answer.
  3. Descriptive analytics
  4. Predictive analytics

Explanation

Prescriptive analytics moves beyond forecasting to recommend the action that optimizes a stated objective, here expected total claim cost. Estimating future claim outcomes is the predictive component, but the recommendation step makes the overall solution prescriptive.

Question 5Using Data Analytics to Strengthen the Insurance Value Chain

An insurer writes $20 million of earned premium and incurs $14 million of losses and loss adjustment expenses in the same period. Its loss ratio is:

  1. 143%
  2. 70% Correct answer.
  3. 60%
  4. 30%

Explanation

Loss ratio equals incurred losses and loss adjustment expenses divided by earned premium, so $14 million divided by $20 million is 70%. It measures the share of premium consumed by claims before underwriting expenses are added.

Question 6Using Data Analytics to Strengthen the Insurance Value Chain

Why do property and casualty insurers apply predictive triage models at first notice of loss?

  1. To route each claim to a path matched to its predicted complexity Correct answer.
  2. To calculate the reinsurance commission that the cedent owes on ceded premium
  3. To set the renewal premium that will be charged to the affected policyholder
  4. To determine the statutory reserve balance that the regulator will require

Explanation

Triage models predict characteristics such as severity, litigation risk, and fraud propensity so claims are assigned early to fast-track, specialist, or investigation units. Matching handling effort to predicted need lowers cost and improves outcomes on complex claims.

Question 7Using Data Analytics to Strengthen the Insurance Value Chain

An insurer has no reliable historical labels identifying fraudulent claims but wants to surface suspicious patterns for investigator review. Which technique fits best?

  1. A linear regression of claim severity on policy tenure and prior claims
  2. A supervised classifier trained on confirmed fraud outcomes from prior years
  3. Unsupervised anomaly detection and clustering of claim characteristics Correct answer.
  4. A randomized controlled trial comparing payment amounts across claim groups

Explanation

When labels are unavailable, unsupervised methods flag claims that deviate from normal patterns and can group them into clusters for investigation. Supervised models require the labeled target that this insurer does not have.

Question 8Using Data Analytics to Strengthen the Insurance Value Chain

Usage-based auto insurance programs depend most directly on which data source?

  1. Vehicle telematics feeds Correct answer.
  2. Credit bureau attributes
  3. Motor vehicle records
  4. Catastrophe risk models

Explanation

Usage-based insurance prices or discounts on measured driving behavior such as miles driven, braking, and time of day, which telematics devices or smartphone apps capture. The other sources inform rating but do not measure actual usage.

Question 9Using Data Analytics to Strengthen the Insurance Value Chain

In a generalized linear model used to price homeowners insurance, what does the estimated coefficient for a rating factor represent?

  1. The factor's effect on the target relative to its base level, all else equal Correct answer.
  2. The likelihood that the factor will be challenged by a regulator during review
  3. The average number of claims observed for every policy in the portfolio
  4. The percentage of policies in the book of business that carry that factor level

Explanation

A GLM coefficient is estimated conditional on the other terms in the model, so it captures the factor's marginal effect relative to the reference level on the model's link scale. Multiplying factor relativities together produces the predicted premium for a risk profile.

Question 10Using Data Analytics to Strengthen the Insurance Value Chain

A territory has 200 exposures and an observed loss ratio of 85%. The complement of credibility is the statewide loss ratio of 65%. Using limited fluctuation credibility, where the exposure needed for full credibility is 1,082, what is the credibility-weighted loss ratio closest to?

  1. 74% Correct answer.
  2. 78%
  3. 66%
  4. 85%

Explanation

Credibility Z equals the square root of 200 divided by 1,082, which is about 0.43. The weighted estimate is 0.43 times 85% plus 0.57 times 65%, giving roughly 73.6% and rounding to 74%.

Question 11Using Data Analytics to Strengthen the Insurance Value Chain

Which description best characterizes a data lake as used by an insurer?

  1. A normalized relational store that holds only fully curated, modeled reporting tables for finance
  2. A repository storing raw data in native formats, with schema applied on read Correct answer.
  3. A messaging service that transmits claim events between core systems in real time
  4. A reporting layer that contains only summarized financial figures for executive scorecards

Explanation

A data lake retains source data in its native form, including text, JSON, and sensor feeds, and defers structure until the point of use. That flexibility supports exploratory analytics and machine learning that a purely modeled warehouse may not accommodate.

Question 12Using Data Analytics to Strengthen the Insurance Value Chain

Why do property insurers geocode their commercial exposures to latitude and longitude?

  1. To aggregate exposure by location and assess catastrophe accumulation Correct answer.
  2. To compute the agent commission rate payable on each individual policy
  3. To satisfy the accounting standards that govern premium recognition timing
  4. To determine the creditworthiness of each commercial tenant before binding coverage

Explanation

Catastrophe losses are driven by geographic concentration, so exposures must be mapped to hazard footprints and summed within event footprints. That aggregation reveals whether a single hurricane or earthquake could breach a reinsurance limit.

Question 13Using Data Analytics to Strengthen the Insurance Value Chain

In the chain ladder reserving method, how are age-to-age development factors used?

  1. They convert paid losses into incurred losses for statutory financial reporting purposes
  2. They project each accident year's cumulative losses to an estimated ultimate Correct answer.
  3. They estimate the number of claims that will be reported in future accident years
  4. They are applied to premium by coverage to establish the expense ratio for reporting

Explanation

Chain ladder computes ratios of cumulative losses at successive maturities and applies them down the development triangle to estimate ultimate losses for each accident year. The gap between ultimate and reported losses produces the IBNR estimate.

Question 14Using Data Analytics to Strengthen the Insurance Value Chain

Which metric best measures the underwriting profitability of a property and casualty portfolio?

  1. Average claim frequency
  2. Quote-to-bind ratio
  3. Retention ratio
  4. Combined ratio Correct answer.

Explanation

The combined ratio adds the loss and loss adjustment expense ratio to the underwriting expense ratio, so a value below 100% signals an underwriting profit. The other metrics describe volume, sales efficiency, or claim frequency rather than overall underwriting profitability.

Question 15Using Data Analytics to Strengthen the Insurance Value Chain

A personal lines insurer wants to decide which policyholders should receive an expensive retention offer. Which approach best targets the spend?

  1. Offer it to the policyholders who have maintained the longest tenure with the insurer
  2. Rank by expected lifetime value times predicted lapse probability Correct answer.
  3. Offer it to every policyholder in the state regardless of segment or risk
  4. Offer it to the policyholders who currently pay the lowest annual premium

Explanation

Expected value at risk combines the economic value of keeping a customer with the probability that the customer would otherwise leave, so spending flows to cases with the greatest expected loss. Tenure or premium alone ignore both churn risk and margin at stake.

Question 16Using Data Analytics to Strengthen the Insurance Value Chain

An insurer runs two versions of a digital quote funnel with traffic randomly assigned. Which result best supports rolling out the new version?

  1. A statistically significant bind-rate lift with no offsetting cost increase Correct answer.
  2. The new version received a higher total number of visits during the test period
  3. The new version was reviewed favorably by the marketing leadership team
  4. The new version generated a lower average quoted premium across test traffic

Explanation

Random assignment makes the comparison causal, and statistical significance separates a real lift from random noise. A rollout decision should also confirm that the gain is not offset by higher acquisition cost or worse downstream loss experience.

Question 17Using Data Analytics to Strengthen the Insurance Value Chain

A claims fraud model classifies 98% of training records correctly but only 61% of records in a holdout sample. Which action addresses the most likely cause?

  1. Reduce complexity with regularization and fewer features, then re-validate Correct answer.
  2. Lower the classification threshold so that a larger share of claims is flagged
  3. Add more derived features so the model can capture additional patterns in the training data
  4. Report the training accuracy as the model's expected performance in production scoring

Explanation

A wide gap between training and holdout performance is the signature of overfitting, where the model memorizes noise specific to the training sample. Regularization, feature reduction, and cross-validated selection reduce variance and produce performance that generalizes.

Question 18Using Data Analytics to Strengthen the Insurance Value Chain

Which of the following is generally treated as personally identifiable information?

  1. A vehicle's make, model, and model year as recorded at binding
  2. An industry classification code assigned to a commercial risk
  3. A policyholder's driver's license number Correct answer.
  4. A statewide average loss ratio for personal auto

Explanation

PII is data that can identify a specific individual, directly or in combination with other data, and a government-issued identifier such as a driver's license number qualifies. Aggregated statistics, industry codes, and generic vehicle attributes do not identify a person on their own.

Question 19Using Data Analytics to Strengthen the Insurance Value Chain

What is the primary responsibility of a data steward in an insurer's analytics program?

  1. Writing and deploying production code for the claims administration system
  2. Auditing the insurer's financial statements before regulatory filing each year
  3. Owning definitions, quality rules, and appropriate use for a data domain Correct answer.
  4. Approving the insurer's annual capital and operating budget

Explanation

Data stewards provide business-side accountability for an assigned data domain, resolving definition conflicts, setting quality thresholds, and deciding who may use the data and for what purpose. They work alongside IT, which implements technical controls.

Question 20Using Data Analytics to Strengthen the Insurance Value Chain

An insurer removes a protected characteristic from a rating model, but a retained variable reproduces nearly the same effect on outcomes across protected groups. Which control best addresses this risk?

  1. Increase model complexity so that other variables dilute the proxy effect
  2. Confirm that the protected field is absent from the modeling dataset and stop there
  3. Test outcomes for disparate impact and evaluate less discriminatory alternatives Correct answer.
  4. Report only the aggregate model accuracy to regulators during filing review

Explanation

Removing a field does not remove its influence when correlated variables act as proxies, so impact testing on model outputs is required. Where disparities appear, the insurer should examine whether a less discriminatory variable or specification achieves similar predictive value.

Question 21Using Data Analytics to Strengthen the Insurance Value Chain

Which technique is best suited to converting free-text adjuster notes into standardized loss-cause categories?

  1. A time-series forecast of monthly reported claim counts
  2. A geographic information system overlay of exposure points
  3. Cluster analysis of premium leakage across rating variables
  4. Natural language processing with text classification Correct answer.

Explanation

Text classification models learn from labeled notes to assign categories such as water damage, fire, or collision to unstructured narrative fields. Forecasting, numeric clustering, and spatial overlays operate on entirely different data types.

Question 22Using Data Analytics to Strengthen the Insurance Value Chain

How can analytics most directly improve an insurer's subrogation results?

  1. By eliminating the need to document individual claim payments in the file
  2. By scoring claims for recovery potential so promising files are referred early Correct answer.
  3. By shortening the statutory notice period for reporting a new claim
  4. By increasing the premium charged to policyholders who have filed prior claims

Explanation

Subrogation value is often lost through late identification, so a model that flags third-party liability and recovery likelihood at first notice or at closure lets specialists pursue the strongest opportunities. Premium adjustments and notice periods are not recovery levers.

Question 23Using Data Analytics to Strengthen the Insurance Value Chain

Which metric most directly measures an agent's effectiveness at turning quoted prospects into bound policies?

  1. Loss ratio
  2. Average premium per policy
  3. Policies in force
  4. Quote-to-bind ratio Correct answer.

Explanation

The quote-to-bind ratio divides bound policies by quotes issued, isolating conversion from volume effects. Premium per policy, in-force counts, and loss ratio describe price level, book size, and profitability instead.

Question 24Using Data Analytics to Strengthen the Insurance Value Chain

After 18 months in production, a pricing model's predictions remain well calibrated on the input distributions, but the historical relationship between its predictors and observed losses has weakened noticeably. Which phenomenon is this?

  1. Sample selection bias
  2. Survivorship bias
  3. Concept drift Correct answer.
  4. Overfitting

Explanation

Concept drift occurs when the underlying relationship between inputs and the target changes, so a model fitted to old patterns no longer predicts well even though inputs look familiar. Covariate drift, by contrast, is a shift in the distribution of the input variables themselves.

Question 25Using Data Analytics to Strengthen the Insurance Value Chain

An insurer wants to know whether a new claims-handling workflow actually reduced average severity. Which design best supports a causal conclusion?

  1. Compare pilot offices with matched non-adopting offices before and after rollout Correct answer.
  2. Ask claims managers whether they believe severity improved after the rollout
  3. Compare average severity before and after rollout across all offices
  4. Compare post-rollout severity between adopting offices and offices that declined

Explanation

A difference-in-differences design using matched comparison offices controls for common shocks such as inflation or weather that would otherwise be mistaken for a workflow effect. Single-group before-and-after and post-only comparisons confound the treatment with concurrent trends and pre-existing differences.

Question 26Using Data Analytics to Strengthen the Insurance Value Chain

Which metric is calculated as incurred losses divided by earned premium?

  1. Expense ratio
  2. Combined ratio
  3. Retention ratio
  4. Loss ratio Correct answer.

Explanation

The loss ratio equals incurred losses divided by earned premium and is a core measure of underwriting performance. The expense ratio measures operating expenses relative to premium, and the combined ratio adds the loss and expense ratios. Retention ratio measures renewal persistence rather than loss cost.

Question 27Using Data Analytics to Strengthen the Insurance Value Chain

What is the primary purpose of data lineage in an insurance analytics environment?

  1. Reduce the storage cost of archived claim files and backup tapes
  2. Trace data origin and transformations for reports and models Correct answer.
  3. Increase the predictive accuracy of every deployed model automatically
  4. Automate the filing of rate changes with insurance regulators

Explanation

Data lineage documents where data comes from, how it is transformed, and where it flows in reports and models. This supports auditability, impact analysis, and regulatory review. It does not by itself improve model accuracy or automate rate filings.

Question 28Using Data Analytics to Strengthen the Insurance Value Chain

Why do insurers use customer lifetime value models in marketing?

  1. To set reinsurance attachment points for catastrophe cover
  2. To estimate the average severity of bodily injury claim payments
  3. To calculate statutory reserves for all unpaid claim obligations
  4. To prioritize marketing spending by customer segment Correct answer.

Explanation

Customer lifetime value estimates the net value a customer is expected to generate over the relationship. Insurers use it to decide where acquisition and retention spending is most worthwhile across segments. It is not a reserving or catastrophe reinsurance calculation.

Question 29Using Data Analytics to Strengthen the Insurance Value Chain

Which of the following is a standard data quality dimension?

  1. Completeness Correct answer.
  2. Development factor
  3. Credibility
  4. Price elasticity

Explanation

Completeness is a standard data quality dimension that measures whether required values are present. Credibility and development factors are actuarial concepts, and price elasticity is an economic concept. Data quality also includes accuracy, consistency, and timeliness.

Question 30Using Data Analytics to Strengthen the Insurance Value Chain

What does a policyholder churn model typically predict?

  1. The regulatory capital required for a line of business
  2. The expected number of claims a policy will file
  3. The commission rate paid to an agent
  4. The probability that a policy will not be renewed Correct answer.

Explanation

A churn model predicts the likelihood that a policyholder will leave, typically by not renewing or canceling. Insurers use it to target retention offers and service outreach. It does not predict claim counts, capital, or commission rates.

Question 31Using Data Analytics to Strengthen the Insurance Value Chain

What is the primary role of a data catalog in an insurance analytics team?

  1. To calculate loss ratios for each rating territory
  2. To store and archive only unstructured claim photographs
  3. To help users find and understand available data assets Correct answer.
  4. To replace actuarial reserving and capital software

Explanation

A data catalog is an inventory that describes available datasets, their definitions, owners, and locations. It helps analysts find and correctly use data without relying on informal knowledge. It is not a storage repository for only photos or a replacement for reserving software.

Question 32Using Data Analytics to Strengthen the Insurance Value Chain

Which item is an example of unstructured data commonly used by insurers?

  1. A list of claim reserve amounts by claim number
  2. A table of premium payments by policy month
  3. Adjuster notes recorded as free text Correct answer.
  4. A spreadsheet of agent commission rates

Explanation

Unstructured data lacks a predefined tabular format; free-text adjuster notes are a common example. Tables of payments, reserves, and commissions are structured because they fit defined rows and columns. Text analytics and NLP can turn notes into usable categories.

Question 33Using Data Analytics to Strengthen the Insurance Value Chain

In ratemaking, what does exposure represent?

  1. The interest rate earned on invested premium
  2. The expense associated with issuing a policy
  3. A unit of risk exposed to loss, such as payroll Correct answer.
  4. The amount of reinsurance purchased by the insurer

Explanation

Exposure is the unit of risk being rated, such as car-years for auto or payroll for workers' compensation. It provides the denominator for frequency and loss cost calculations. Reinsurance amounts, expenses, and investment rates are not exposure units.

Question 34Using Data Analytics to Strengthen the Insurance Value Chain

An insurer wants to group agents by profitability, growth, and retention to decide which agencies receive additional support. Which analytic technique is most appropriate?

  1. Chain ladder development
  2. Randomized controlled trial on agent commissions
  3. Cluster analysis on agent performance metrics Correct answer.
  4. Credibility weighting of loss ratios

Explanation

Cluster analysis groups agents with similar performance profiles without requiring predefined labels. That helps an insurer tailor support, compensation, or training to distinct agency segments. Chain ladder and credibility are reserving tools, and an RCT tests an intervention rather than segments the population.

Question 35Using Data Analytics to Strengthen the Insurance Value Chain

A workers' compensation insurer wants to flag employers whose claim frequency is likely to rise before renewal. Which model output is most directly useful?

  1. The reinsurance commission rate for the treaty
  2. The insurer's expense ratio for the state
  3. The employer's expected claim severity for the next policy year
  4. Predicted claim frequency trend versus industry peers Correct answer.

Explanation

A frequency trend model estimates whether claim counts are rising for an employer relative to its industry or peers. That output can prioritize renewal review, safety outreach, or pricing action. Severity, state expense ratio, and reinsurance commission do not directly flag rising claim frequency.

Question 36Using Data Analytics to Strengthen the Insurance Value Chain

An insurer combines forecast hail data with policy location data to pre-position claims adjusters before a storm. This practice is best described as:

  1. Testing a new rating factor for catastrophe risk
  2. Using analytics to deploy claims resources Correct answer.
  3. Reserving for incurred but not reported claims
  4. Calculating the combined ratio for a territory

Explanation

Combining forecast weather with policy locations lets an insurer anticipate where claims will arise and position adjusters or vendors. This is an operational use of predictive analytics in claims, not reserving or ratemaking. The goal is faster response and better resource allocation.

Question 37Using Data Analytics to Strengthen the Insurance Value Chain

What is the main pricing advantage of separating expected losses into claim frequency and average severity?

  1. It removes the need to collect exposure data
  2. It eliminates the need for all reinsurance purchases
  3. It guarantees actual losses will equal expected losses
  4. It lets rating factors target frequency and severity Correct answer.

Explanation

Frequency is claims per exposure, and severity is cost per claim; their product gives expected loss cost. Separating them lets pricing target factors that affect claim counts differently from those that affect claim costs. It does not remove exposure data, guarantee results, or eliminate reinsurance.

Question 38Using Data Analytics to Strengthen the Insurance Value Chain

A commercial property underwriter reviews a loss run showing one large fire loss at a single warehouse. Which analytic question is most important before offering terms?

  1. How does this loss affect aggregate risk at nearby locations? Correct answer.
  2. What is the policyholder's average claim frequency for auto liability?
  3. What commission rate did the broker earn last year?
  4. How many quotes did the insurer issue in that industry?

Explanation

A large single loss can change the insurer's aggregate exposure if other insured locations are nearby or share catastrophe risk. Underwriters use accumulation analysis to assess reinsurance needs and capacity limits. Broker commission, auto frequency, and quote volume do not address property concentration.

Question 39Using Data Analytics to Strengthen the Insurance Value Chain

An insurer wants to estimate how quoted premium changes affect the probability that a prospect buys a policy, while controlling for competitor prices. Which method is most appropriate?

  1. Regression on quote-level price data Correct answer.
  2. Chain ladder reserving of paid claim amounts
  3. Bornhuetter-Ferguson reserving of losses
  4. Cluster analysis of claim severity patterns

Explanation

Price elasticity is estimated by modeling purchase outcomes against the quoted premium while controlling for competitor prices and other quote characteristics. Regression on quote-level data is suited to that task. Reserving methods and severity clustering do not estimate demand sensitivity.

Question 40Using Data Analytics to Strengthen the Insurance Value Chain

To detect potential overpayment in auto bodily injury claims, which analytic approach is most direct?

  1. Forecasting catastrophe losses with a vendor model
  2. Estimating price elasticity from quote data
  3. Grouping agents by retention rates
  4. Comparing line items with coverage rules and peers Correct answer.

Explanation

Claims leakage detection compares billed or paid line items against coverage rules, fee schedules, and peer payment patterns to flag outliers. That directs review toward potential overpayments. Catastrophe forecasting, agent retention, and price elasticity do not examine claim line-item appropriateness.

Question 41Using Data Analytics to Strengthen the Insurance Value Chain

A health insurer finds that a provider group has unusually high readmission rates after certain procedures. What is the most appropriate analytic follow-up?

  1. Remove the provider from the network immediately
  2. Increase the provider's capitation rate without review
  3. Compare the provider's case mix and outcomes with peers Correct answer.
  4. Recalculate the insurer's loss ratio using chain ladder

Explanation

Provider profiling compares outcomes after adjusting for case mix, severity, and patient characteristics. An unusually high readmission rate may reflect patient mix rather than poor care. Immediate network removal or rate increases without review would be premature.

Question 42Using Data Analytics to Strengthen the Insurance Value Chain

How can predictive analytics most directly improve a workers' compensation premium audit program?

  1. By determining the number of claims adjusters needed for catastrophe duty
  2. By calculating the insurer's investment income
  3. By setting the discount rate for loss reserves
  4. By ranking policies by underreported payroll risk Correct answer.

Explanation

Premium audit analytics can rank policies by the likelihood that reported payroll or sales are understated. Auditors can then focus field or desk audits where expected recovery is highest. Investment income, reserve discount rates, and catastrophe staffing are unrelated to audit selection.

Question 43Using Data Analytics to Strengthen the Insurance Value Chain

An insurer wants to use thousands of free-text customer complaints to identify recurring service failures. Which analytic step is most appropriate first?

  1. Apply chain ladder to the complaint volume
  2. Calculate the combined ratio for each complaint
  3. Run a price elasticity regression on complaint counts
  4. Classify complaint text into reason categories with NLP Correct answer.

Explanation

Natural language processing can convert free-text complaints into reason categories for counting and trend analysis. That reveals recurring service failures and product issues. Chain ladder, combined ratio, and price elasticity do not classify text.

Question 44Using Data Analytics to Strengthen the Insurance Value Chain

Why is model explainability particularly important when an insurer submits a new rating model for regulatory review?

  1. It removes the need to validate the model's data quality
  2. It allows the insurer to avoid documenting model assumptions
  3. It guarantees the model will have no prediction error
  4. It helps regulators assess rating factors and outputs Correct answer.

Explanation

Regulators review whether rating factors and model outputs are justified, transparent, and consistent with insurance rules. Explainability helps demonstrate the rationale and supports non-discrimination review. It does not eliminate prediction error, replace data validation, or remove documentation needs.

Question 45Using Data Analytics to Strengthen the Insurance Value Chain

An insurer plans to use image analytics on auto damage photos to estimate repair costs. Which condition is most important for building a reliable model?

  1. A large set of labeled photos with repair costs Correct answer.
  2. Replacing all human appraisers immediately
  3. Using only photos taken at night
  4. Ignoring repair estimates when labeling the photos

Explanation

Supervised image models need many labeled examples that connect damage photos to reliable repair cost outcomes. Without accurate labels, the model cannot learn the relationship between visual damage and repair cost. Human oversight and diverse photo conditions also matter.

Question 46Using Data Analytics to Strengthen the Insurance Value Chain

For which situation is the Bornhuetter-Ferguson reserving method generally preferred over the chain ladder method?

  1. A line of business with no expected loss ratio available
  2. A portfolio where all claims are paid immediately
  3. An immature year with little credible loss experience Correct answer.
  4. A fully mature accident year with stable development patterns

Explanation

Bornhuetter-Ferguson blends an expected loss ratio with reported development, which is useful when an accident year is immature and observed losses are not yet credible. Chain ladder relies more heavily on observed development and is less stable in that situation. BF still requires an expected loss ratio assumption.

Question 47Using Data Analytics to Strengthen the Insurance Value Chain

An insurer wants to optimize its catastrophe reinsurance program subject to a capital constraint. Which analytic approach is most appropriate?

  1. Simulate catastrophe losses across reinsurance structures Correct answer.
  2. Use chain ladder to project paid claims by accident year
  3. Classify customer complaints with natural language processing
  4. Estimate the loss ratio for each agent

Explanation

Reinsurance optimization compares alternative retentions, limits, and layers using simulated catastrophe losses and risk-adjusted capital. That lets the insurer evaluate cost versus capital relief under a constraint. Agent loss ratios, complaint text, and chain ladder do not compare reinsurance structures.

Question 48Using Data Analytics to Strengthen the Insurance Value Chain

An insurer offers a telematics discount to policyholders who volunteer to enroll, and wants to estimate the discount's causal effect on retention. Which concern is most important?

  1. Enrollees may differ systematically from non-enrollees Correct answer.
  2. Retention is not measurable for insurance policies
  3. The discount cannot be analyzed without chain ladder reserving
  4. Telematics data cannot be linked to policy records

Explanation

Voluntary enrollment means participants may differ from nonparticipants in ways that also affect retention, creating self-selection bias. A causal estimate should use methods such as propensity score matching or instrumental variables. Simply comparing retention between enrollees and non-enrollees can confound the discount effect.

Question 49Using Data Analytics to Strengthen the Insurance Value Chain

An insurer must decide how often to validate each predictive model. Which factor most directly justifies more frequent validation?

  1. The model drives pricing, claims, or capital decisions Correct answer.
  2. The model was built using a popular programming language
  3. The model's output is displayed on a dashboard
  4. The model has a large number of input variables

Explanation

Model risk tiering uses materiality: models that materially affect pricing, claims, capital, or financial reporting warrant more frequent validation. Programming language, dashboard display, and variable count are not the primary drivers of validation frequency. Higher-impact models also need stronger documentation and controls.

Question 50Using Data Analytics to Strengthen the Insurance Value Chain

Before sharing a claims dataset with a research partner, an insurer applies k-anonymity. What does this technique aim to ensure?

  1. Each record matches at least k-1 others on quasi-identifiers Correct answer.
  2. Every claim amount is encrypted with a different key
  3. All direct identifiers are removed but quasi-identifiers remain unchanged
  4. Only aggregated loss ratios are reported

Explanation

K-anonymity generalizes or suppresses quasi-identifiers so each record is indistinguishable from at least k-1 others. That reduces the risk that a person can be re-identified by combining attributes such as age, ZIP code, and gender. Removing only direct identifiers is insufficient because quasi-identifiers can still enable linkage.