12. Portfolio Analytics and Performance Management

Introduction and Context

Insurance underwriting is ultimately a portfolio activity. An underwriter may approve or decline individual risks, but the financial outcome of the insurer depends on the combined behaviour of thousands of policies, claims, distribution channels, territories, products and customer groups. A portfolio that appears satisfactory at policy level can produce poor results when risks accumulate in one location, claims costs inflate, expenses rise, or renewal retention weakens. Portfolio analytics provides the methods through which insurers convert operational data into underwriting judgement, financial control and strategic action.

Portfolio performance management is broader than calculating a loss ratio. It requires the insurer to understand the frequency and severity of claims, the timing of loss emergence, expense allocation, claims leakage, retention, conversion, service quality and capital consumption. These measures must be examined by meaningful segments and across time. A portfolio can have a strong overall combined ratio while concealing serious deterioration in a particular class, broker channel or policy cohort.

The purpose of analytics is therefore not merely to describe past performance. It is to identify the causes of performance, estimate future outcomes, test resilience under adverse conditions and support decisions about pricing, underwriting authority, claims management, distribution and capital. This chapter develops an integrated approach to these tasks.

Learning Objectives

After studying this chapter, the reader should be able to:

1. Explain the principal measures used in insurance portfolio performance management.

2. Calculate and interpret loss ratio, combined ratio, frequency, severity, retention, conversion and expense indicators.

3. Distinguish reported, paid, incurred and ultimate claims measures.

4. Apply segmentation, cohort analysis, dashboards and stress testing to underwriting portfolios.

5. Recognise claims development, leakage and service outcomes as drivers of profitability.

6. Construct a practical profitability model for a product or portfolio.

7. Evaluate governance, data, ethical and technological issues in portfolio analytics.

8. Recommend managerial actions based on portfolio evidence rather than isolated indicators.

The Portfolio as an Analytical Unit

A portfolio is a defined collection of insurance exposures managed for a common purpose. It may consist of a line of business, such as motor or property; a product; a territory; a broker book; a customer segment; or a cohort of policies written during a particular period. The selection of the unit matters because different units answer different questions. Accident-year analysis is useful for understanding when losses occurred, whereas underwriting-year analysis links premium and claims to the policies written during the same underwriting period.

Portfolio analytics begins with a clear performance question. For example:

• Is the motor portfolio adequately priced?
• Which distribution channels produce profitable customers?
• Are claims costs deteriorating because of frequency, severity or settlement practice?
• Does a high renewal rate compensate for weak margins?
• Can the portfolio withstand inflation, catastrophe accumulation or adverse claims development?

A sound analysis specifies the observation period, exposure basis, inclusion rules, currency, valuation date and segmentation variables. Without these controls, ratios may be mathematically correct but economically misleading.

Core Terminology

Written premium is premium recorded for policies issued during a period. Earned premium is the portion of written premium relating to the insurance protection provided during that period. Because coverage often extends beyond the accounting period, written and earned premium differ.

Reported claims are claims notified to the insurer. Paid claims are amounts already settled. Incurred claims generally comprise paid claims plus outstanding case reserves. Ultimate claims are the estimated final cost after all claims have been reported, developed and settled. The difference between incurred and ultimate claims is particularly important in long-tail business.

Exposure measures the quantity of risk insured. Depending on the class, it may be policy-years, vehicle-years, property values, employee-years, insured days or another appropriate base. Premium alone is not always a reliable measure of portfolio volume, especially when rates change substantially.

Gross figures are before reinsurance, while net figures reflect reinsurance recoveries, ceded premiums and related costs. Both perspectives are necessary: gross performance reflects underwriting quality, whereas net performance reflects the insurer’s retained economic result.

The Performance Measurement Architecture

Frequency and Severity

Claims frequency measures how often claims occur relative to exposure:

\[

\text{Claims Frequency} = \frac{\text{Number of Claims}}{\text{Exposure Units}}

\]

Severity measures the average cost of a claim:

\[

\text{Average Severity} = \frac{\text{Incurred or Paid Claims}}{\text{Number of Claims}}

\]

Expected claims cost per exposure can be expressed as:

\[

\text{Pure Premium} = \text{Frequency} \times \text{Severity}

\]

This decomposition is analytically valuable. A rising loss ratio may result from more claims, larger claims, or both. Frequency may increase because of fraud, poor risk selection, changing usage or external hazards. Severity may rise because of medical inflation, repair costs, litigation, social inflation, supply-chain constraints or claims handling decisions.

The selection of numerator is important. Paid severity is useful for settled claims but may understate the eventual cost of open claims. Incurred severity provides a more current estimate, subject to reserve uncertainty.

Loss Ratio and Combined Ratio

The loss ratio is commonly calculated as:

\[

\text{Loss Ratio} = \frac{\text{Incurred Claims and Related Loss Adjustment Expenses}}{\text{Earned Premium}} \times 100

\]

Some organisations report claims only, while others include allocated and unallocated loss adjustment expenses. The definition should always be stated.

The expense ratio is:

\[

\text{Expense Ratio} = \frac{\text{Underwriting Expenses}}{\text{Earned Premium}} \times 100

\]

The combined ratio is:

\[

\text{Combined Ratio} = \text{Loss Ratio} + \text{Expense Ratio}

\]

A combined ratio below 100 per cent indicates an underwriting profit before investment income and certain other items. It does not necessarily mean that the insurer is economically successful. Capital costs, reinsurance costs, taxes, catastrophe volatility and the time value of money may still produce an inadequate return.

Ratios must be interpreted with their denominators and valuation dates. A high ratio in a new portfolio may reflect immature earned premium or conservative reserving, while a low ratio may reflect claims not yet reported. A portfolio with a 65 per cent loss ratio and a 40 per cent expense ratio has a combined ratio of 105 per cent, even though the loss ratio alone appears acceptable.

Retention, Conversion and Growth

Customer retention measures the proportion of customers or policies renewed. A basic policy retention rate is:

\[

\text{Retention Rate} = \frac{\text{Policies Renewed}}{\text{Policies Eligible for Renewal}} \times 100

\]

Premium retention may be more informative where policy sizes vary significantly. Retention should be analysed alongside profitability. Retaining unprofitable business may destroy value, while losing profitable customers may indicate pricing, service or distribution problems.

Conversion measures the proportion of quotations or leads that become policies:

\[

\text{Conversion Rate} = \frac{\text{Bound Policies}}{\text{Eligible Quotes or Leads}} \times 100

\]

High conversion is not automatically desirable. Aggressive discounts, weak underwriting controls or selective quotation practices may increase conversion while worsening loss ratios. The useful question is whether conversion produces profitable growth at an acceptable level of risk.

Growth should therefore be decomposed into rate change, exposure change, mix change, retention, new business and cancellation. Nominal premium growth can conceal a reduction in exposure or a deterioration in technical price adequacy.

Development and Reserve Performance

Claims development describes how estimates change from initial notification to final settlement. Development triangles arrange claims by accident or underwriting period and development age. Common measures include paid development, incurred development, reported claim counts and case reserve development.

Important concepts include:

• Incurred but not reported (IBNR): expected claims that have occurred but are not yet reported.
• Case reserve adequacy: the extent to which individual claim reserves reflect expected settlement costs.
• Development factor: the estimated relationship between claims at one development age and a later age.
• Ultimate loss estimate: the projected final cost of a claims cohort.

Methods such as the chain-ladder approach use historical development patterns, while the Bornhuetter–Ferguson approach combines an expected loss ratio with emerging claims experience. These methods require judgement where portfolios are immature, changing rapidly or affected by unusual events. Reserve indications should be compared with actual emergence through back-testing.

A development result may arise from genuine claims deterioration, delayed reporting, changes in case reserving, settlement inflation or operational disruption. Management should avoid treating every reserve strengthening as an underwriting failure, but repeated adverse development is a strong signal that pricing, reserving or claims controls require review.

Segmentation: From Portfolio Averages to Actionable Insight

A portfolio average is a starting point, not a decision. Segmentation divides the book into groups with sufficiently different risk, behaviour or economics to justify separate analysis. Relevant variables may include geography, product, customer tenure, occupation, vehicle type, property construction, sum insured, broker, digital channel, policy size, deductible, claims history and renewal status.

Segmentation must balance statistical credibility and managerial usefulness. Excessively narrow segments produce unstable results and privacy risks. Excessively broad segments conceal adverse selection. A segment should generally have a clear business rationale, adequate data quality, sufficient exposure and an identifiable action associated with its performance.

A useful segmentation process is:

1. Define the business question and outcome measure.

2. Validate exposure, premium and claims data.

3. Establish logical and regulatory permissible variables.

4. Examine frequency, severity, loss ratio, expenses and retention by segment.

5. Test whether differences persist after controlling for mix and development.

6. Identify underwriting, pricing, claims or service actions.

7. Monitor the segment after intervention.

Interaction effects are often important. A territory may be profitable overall but unprofitable for a particular vehicle type or distribution channel. Segment analysis should therefore examine combinations cautiously and apply credibility techniques where data are sparse.

Cohort Analysis and the Underwriting Cycle

A cohort is a group of policies sharing a common origin, such as inception quarter, underwriting year, acquisition campaign or renewal month. Cohort analysis reveals how performance evolves as business matures. It can distinguish whether deterioration is concentrated in recently written business or reflects a broad portfolio problem.

For example, a claims ratio may appear stable at portfolio level while successive quarterly cohorts show progressively higher ultimate loss estimates. This pattern can indicate underpricing, relaxed acceptance standards or a change in external risk. Conversely, a recent cohort may initially appear poor because it has not yet earned all premium or because early claims are reported faster than historical claims.

Cohort analysis should track:

• written and earned premium;
• exposure and policy count;
• claim frequency and severity;
• incurred and ultimate loss ratio;
• acquisition and servicing cost;
• retention and cancellation;
• channel and product mix; and
• development from initial to mature estimates.

The underwriting cycle introduces changing market conditions. In a soft market, capacity and competition may reduce prices and encourage broader terms. In a hard market, prices and underwriting discipline generally strengthen. Portfolio managers should separate cycle effects from structural changes such as technology, regulation, climate exposure or claims inflation. A profitable year may reflect favourable market pricing rather than superior underwriting capability.

Profitability Models and Value-Based Management

A practical profitability model connects revenue, expected claims, expenses, capital and customer behaviour. At policy or segment level, a simplified expected result may be represented as:

\[

\text{Expected Underwriting Margin} = \text{Earned Premium} - \text{Expected Claims} - \text{Operating Expenses} - \text{Reinsurance Cost}

\]

A more complete model may include investment income, cost of capital, taxes, acquisition costs, cancellation effects and the expected value of future renewals. Customer lifetime value can be estimated by combining expected margin, retention probability, servicing cost and the time value of money. Such models should not encourage indiscriminate rejection of high-cost customers; rather, they should support fair pricing, appropriate service design and sustainable risk selection.

Profitability attribution is a managerial challenge. Underwriting teams may be assessed on risk-adjusted technical margin, claims teams on ultimate claims and leakage, distribution teams on profitable conversion and retention, and senior management on portfolio return and capital efficiency. The attribution system should recognise dependencies. A claims outcome may reflect the original underwriting decision, policy wording, vendor quality and inflation, not merely claims department performance.

Claims Leakage and Service Outcomes

Claims leakage is the value lost through avoidable overpayment, underpayment, process failure, fraud, inappropriate settlement, missed recoveries or failure to apply policy terms correctly. It is not simply the same as fraud. Leakage may arise from duplicate payments, incorrect reserves, weak repair controls, delayed subrogation, poor vendor management or inconsistent interpretation of coverage.

Analytics can identify unusual payment patterns, excessive settlement durations, frequent reopened claims, high litigation rates, inconsistent reserving or variance between adjusters. However, controls must distinguish legitimate complexity from poor practice. Overly aggressive leakage reduction can create unfair claim denials, complaints and regulatory action.

Service outcomes should be measured alongside financial outcomes. Relevant indicators include first-notification acknowledgement time, average settlement time, complaint rate, claims abandonment, contact resolution, customer satisfaction and vulnerable-customer outcomes. A low claims cost achieved through unreasonable delay is not sound performance management. Quality, fairness and timeliness are part of the claims value proposition.

Dashboards, Data and Decision Processes

An effective portfolio dashboard is a decision instrument rather than a display of numerous metrics. It should show current performance, trend, target, variance, confidence or uncertainty, and the responsible action owner. A balanced dashboard may include:

• volume: exposure, policies, premium and growth;
• risk: frequency, severity, catastrophe concentration and mix;
• profitability: loss, expense and combined ratios;
• claims: development, leakage, recovery and settlement time;
• customer: retention, conversion, complaints and service outcomes; and
• control: data quality, exceptions, authority breaches and model drift.

Dashboards should permit drill-down from total portfolio to product, geography, channel and cohort. Colour coding without context can mislead, particularly where a small segment has volatile results. Each metric should have a definition, owner, source, refresh frequency and escalation threshold.

Data governance is foundational. Common problems include duplicate policy records, inconsistent claim status, missing exposure data, mismatched policy and claim identifiers, changing definitions and incomplete expense allocation. A data dictionary and controlled metric catalogue prevent different departments from reporting incompatible versions of the same ratio.

Stress Testing and Scenario Analysis

Historical results do not reveal the full range of possible outcomes. Stress testing examines the effect of adverse but plausible changes. Scenarios may include a rise in claim frequency, higher repair or medical inflation, delayed settlement, catastrophe accumulation, reinsurance failure, a sudden fall in retention, cyber interruption or regulatory change.

A simple scenario model might increase frequency by 15 per cent, severity by 10 per cent and expenses by 5 per cent, then estimate the effect on the combined ratio and capital requirement. More advanced models use stochastic simulations and dependency assumptions. Scenario analysis should state whether shocks are gross or net of reinsurance, whether management actions are allowed and how quickly assumptions apply.

Reverse stress testing begins with a failure condition, such as a combined ratio above a specified threshold or breach of a capital limit, and asks what combination of events could produce it. This supports contingency planning and risk appetite decisions.

Stress tests are not forecasts. Their value lies in revealing sensitivity, concentration and management options. Results should lead to actions such as adjusting limits, purchasing reinsurance, revising underwriting rules, strengthening reserves, diversifying channels or increasing operational capacity.

Technology, Automation and Artificial Intelligence

Modern insurers use data warehouses, cloud platforms, application programming interfaces, business-intelligence tools and machine-learning models to analyse portfolio performance. Automated dashboards can identify emerging deterioration faster than periodic manual reports. Predictive models may estimate claim frequency, severity, lapse probability, fraud likelihood or expected lifetime value.

Technology creates material risks. A model trained on biased historical decisions may reproduce unfair outcomes. Data drift can reduce predictive accuracy when customer behaviour, repair costs or legal conditions change. Explainability is particularly important where an automated decision affects access to insurance or claims payment. Human oversight, model validation, monitoring, documentation, access controls and audit trails are essential.

Generative AI may assist with summarising claims files, identifying anomalies or producing management commentary, but it should not be treated as an independent authority. Confidential information must be protected, outputs verified and accountability retained by qualified professionals. Automation should improve consistency and speed without removing the right to review, challenge and appeal.

Governance, Ethics and Regulatory Considerations

Portfolio analytics influences pricing, acceptance, claims settlement and customer treatment. Governance must therefore address not only technical accuracy but also fairness, transparency, privacy and accountability. Sensitive characteristics and proxy variables require careful treatment. A segment that appears profitable may reflect exclusion of vulnerable customers or discriminatory historical practices.

Management should establish clear authority for metric definitions, model approval, data access, exception handling and corrective action. Regulatory reporting must reconcile with internal performance measures. Claims analytics should respect policyholder rights, complaint procedures and applicable conduct standards. Decisions should be explainable enough to support internal challenge, customer communication and supervisory review.

Common mistakes include relying on one ratio, comparing immature and mature cohorts, confusing correlation with causation, ignoring reinsurance effects, treating model output as fact, failing to adjust for rate changes, and rewarding growth without considering ultimate profitability. A sophisticated analytical culture values uncertainty and records the assumptions behind decisions.

Illustrative Case Study: The Meridian Motor Portfolio

Background

Meridian General Insurance, a fictional insurer, writes personal motor business through brokers and a digital channel. Its annual dashboard showed a stable 72 per cent loss ratio and a 28 per cent expense ratio, producing a 100 per cent combined ratio. Management initially concluded that the portfolio was broadly under control.

Situation and Problem

Despite stable headline ratios, cash flow weakened and customer complaints about repair delays increased. The chief underwriting officer requested a deeper review before approving further growth.

Analysis

The analytics team separated the portfolio by underwriting quarter, channel, vehicle age, territory and customer tenure. It found that the most recent four cohorts had materially higher estimated ultimate severity than earlier cohorts. Frequency had increased modestly, but severity had risen sharply because of parts inflation and longer repair cycles.

The digital channel showed a higher conversion rate than brokers, but its new-business claims frequency was also higher. Retention was strongest among customers with low claims experience, while profitable customers were receiving aggressive renewal increases and leaving. A claims review identified delayed subrogation and inconsistent use of preferred repair networks. The total portfolio ratio concealed these effects because older profitable cohorts remained in the denominator.

Stress testing indicated that a further 10 per cent increase in severity would push the combined ratio materially above the acceptable threshold. The analysis also showed that reported claims were understated in the newest cohorts because of normal reporting and settlement lags.

Approach and Strategy

Meridian implemented a coordinated response. It revised vehicle and territory pricing using frequency-severity analysis, introduced tighter referral rules for selected combinations, and reviewed digital-channel acquisition discounts. Claims management renegotiated repair-network standards, automated recovery reminders and introduced weekly monitoring of reopened claims and settlement duration. Renewal decisions were changed from blanket increases to segment-specific actions based on expected profitability and customer value.

The dashboard was redesigned to show accident-year and underwriting-year views, ultimate loss estimates, cohort maturity, service indicators and stress-test results. Governance approval was required before adding a new segment or changing a metric definition.

Outcome

The intervention did not seek immediate volume growth. Instead, Meridian slowed acquisition in the most adverse segments, improved claims recovery controls and redirected capacity toward better-performing risks. Management reported improved visibility into emerging loss costs and reduced reliance on the apparently stable headline ratio. The case does not claim a specific numerical improvement; its significance lies in demonstrating how analytical diagnosis precedes responsible action.

Learning Points

First, portfolio averages can conceal deterioration in new cohorts. Second, conversion and retention must be evaluated with profitability and service quality. Third, claims operations are an important source of underwriting performance. Fourth, stress testing reveals vulnerabilities that historical ratios may not show. Finally, performance management is cross-functional: pricing, distribution, claims, finance, data and governance must act together.

Best-Practice Principles

A mature portfolio analytics function should:

1. Define every metric, denominator, valuation date and scope.

2. Use exposure-based measures and separate frequency from severity.

3. Analyse both gross and net results and reconcile them to financial accounts.

4. Examine cohorts and claims development rather than relying on annual averages.

5. Combine profitability, customer, claims and control indicators.

6. Make every dashboard threshold linked to a decision and accountable owner.

7. Quantify uncertainty and test adverse but plausible scenarios.

8. Validate models, monitor drift and retain human oversight.

9. Investigate causes before imposing corrective action.

10. Protect fairness, privacy, transparency and policyholder rights.

Current Developments and Emerging Directions

Portfolio management is increasingly influenced by climate-related loss volatility, claims inflation, changing mobility patterns, embedded insurance, ecosystem partnerships and real-time data. Usage-based insurance and connected devices can improve exposure measurement but raise consent and privacy questions. Parametric products require analytics focused on trigger reliability and basis risk. Climate scenario analysis is expanding from catastrophe modelling into asset, underwriting and claims strategy.

The direction of travel is toward more frequent, granular and forward-looking monitoring. Nevertheless, better technology does not eliminate the need for underwriting judgement. The central managerial task remains the translation of uncertain evidence into fair, profitable and resilient decisions.

Chapter Summary

Portfolio analytics connects policy-level underwriting and claims activity with enterprise performance. Frequency and severity explain the drivers of claims cost; loss, expense and combined ratios assess underwriting results; retention and conversion reveal customer and distribution behaviour; and development analysis estimates how claims mature toward ultimate cost. Segmentation and cohort analysis prevent portfolio averages from concealing adverse selection or emerging deterioration.

Effective performance management combines profitability models with claims leakage, service outcomes, data governance and stress testing. Dashboards should support decisions rather than merely report information. Technology and AI can improve speed and insight, but require validation, fairness controls, privacy protection and human accountability. The most reliable portfolio decisions are evidence-based, forward-looking, cross-functional and attentive to both financial and customer outcomes.

Key Terms

Accident Year: The year in which an insured loss occurs.

Cohort: A group of policies sharing a common origin or time period.

Combined Ratio: Loss ratio plus expense ratio.

Conversion Rate: The proportion of eligible quotations or leads that become policies.

Earned Premium: Premium relating to insurance protection provided during a period.

Exposure: A measure of the quantity of risk insured.

Expense Ratio: Underwriting expenses divided by earned premium.

Frequency: The number of claims relative to exposure units.

IBNR: Estimated claims incurred but not yet reported.

Incurred Claims: Paid claims plus outstanding case reserves.

Loss Ratio: Incurred claims and specified loss-adjustment costs divided by earned premium.

Pure Premium: Expected claims cost per exposure, often expressed as frequency multiplied by severity.

Retention Rate: The proportion of eligible policies or customers that renew.

Severity: Average cost per claim.

Stress Testing: Examination of portfolio results under adverse but plausible assumptions.

Underwriting Year: The year in which policies are written or incepted.

Ultimate Claims: The projected final cost of claims after development and settlement.

Claims Leakage: Avoidable financial loss arising from overpayment, process failure, missed recovery or related weaknesses.

Review Questions

Short Answer Questions

1. Why can written premium and earned premium differ?

2. Distinguish claims frequency from claims severity.

3. What does a combined ratio below 100 per cent generally indicate?

4. Why is cohort analysis useful in portfolio management?

5. Identify two examples of claims leakage.

Long Answer Questions

1. Explain the construction and interpretation of loss ratio, expense ratio and combined ratio.

2. Discuss how claims development triangles and ultimate loss estimates support reserve and underwriting decisions.

3. Explain the role of segmentation in identifying profitable and unprofitable insurance business.

4. Describe the design features of an effective portfolio performance dashboard.

5. Evaluate the governance, ethical and operational risks of using AI in insurance portfolio analytics.

Analytical / Application Questions

1. A portfolio has a 68 per cent loss ratio and a 36 per cent expense ratio. Interpret the combined ratio and explain why further analysis may still be necessary.

2. A digital channel has a 15 per cent higher conversion rate but materially worse claim frequency than the broker channel. What additional evidence should management examine before expanding the digital channel?

3. A new underwriting cohort shows a poor incurred loss ratio, but claims reporting is known to be delayed. Explain how you would assess whether the deterioration is genuine and what decisions should be postponed or taken cautiously.

22. Multiple Choice Questions with Answers

Question 1

A portfolio has an incurred loss ratio of 64% and an expense ratio of 31%. What is its combined ratio?

A. 33%

B. 64%

C. 95%

D. 105%

Correct Answer: C. 95%

Question 2

Which measure best explains whether a rising claims cost is caused by more claims or larger claims?

A. Retention and conversion

B. Frequency and severity

C. Written and earned premium

D. Gross and net premium

Correct Answer: B. Frequency and severity

Question 3

Why can a new underwriting cohort show an apparently poor incurred loss ratio even when its final result is uncertain?

A. Its claims may not yet be fully reported or developed

B. Its expense ratio is always zero

C. Its written premium is always equal to exposure

D. Its retention rate determines ultimate claims

Correct Answer: A. Its claims may not yet be fully reported or developed

Question 4

Which action is most appropriate when a high conversion segment also has materially adverse claims experience?

A. Expand it immediately because conversion is high

B. Ignore claims because conversion is a customer metric

C. Assess profitable conversion using claims, expenses and retention together

D. Remove all customers from the segment without further analysis

Correct Answer: C. Assess profitable conversion using claims, expenses and retention together

Question 5

IBNR refers to:

A. Premium earned but not collected

B. Claims incurred but not yet reported

C. Expenses incurred but not allocated

D. Policies renewed but not issued

Correct Answer: B. Claims incurred but not yet reported

Question 6

Which dashboard design is most useful for management action?

A. A display of maximum possible metrics without definitions

B. A single annual combined ratio for the whole company

C. Metrics showing trend, target, variance, uncertainty and accountable action owners

D. Only customer satisfaction scores

Correct Answer: C. Metrics showing trend, target, variance, uncertainty and accountable action owners

Question 7

A pure premium is commonly estimated as:

A. Frequency multiplied by severity

B. Retention multiplied by conversion

C. Loss ratio plus expense ratio

D. Written premium divided by earned premium

Correct Answer: A. Frequency multiplied by severity

Question 8

Which is an example of claims leakage?

A. A properly paid claim within policy terms

B. A missed subrogation recovery caused by process failure

C. A correctly calculated premium refund

D. A documented catastrophe reserve

Correct Answer: B. A missed subrogation recovery caused by process failure

Question 9

What is the primary purpose of reverse stress testing?

A. To calculate historical retention

B. To identify what combination of events could produce a defined failure condition

C. To replace all financial reporting

D. To measure only customer satisfaction

Correct Answer: B. To identify what combination of events could produce a defined failure condition

Question 10

Why should AI outputs in underwriting and claims analytics remain subject to human oversight?

A. Models cannot process any insurance data

B. Historical data may contain bias, drift or errors affecting decisions

C. Automation always increases claims leakage

D. Human review eliminates the need for model validation

Correct Answer: B. Historical data may contain bias, drift or errors affecting decisions