Product Governance, Underwriting and Pricing Regulation

Introduction

Insurance products are not merely commercial offerings. They are legally structured promises under which an insurer accepts uncertain risk in return for consideration. Product governance therefore connects contract law, prudential supervision, consumer protection, actuarial science, conduct regulation and corporate accountability. A product may be technically profitable yet unlawful if its exclusions are misleading, its price is discriminatory, its documentation is incomprehensible, or its distribution process fails to assess suitability.

This chapter examines the regulatory control of insurance products across life, health, property, casualty and specialty classes. It uses a principles-based perspective applicable across jurisdictions, while recognising that detailed requirements vary between national insurance statutes, supervisory regulations and market-conduct rules.

Learning Objectives

By the end of this chapter, the reader should be able to:

1. Explain the legal and governance foundations of insurance product regulation.
2. Evaluate product design, approval and lifecycle-monitoring systems.
3. Analyse underwriting classification, actuarial fairness and pricing controls.
4. Assess policy wording, exclusions, suitability and disclosure.
5. Apply governance principles to life, health, property and casualty insurance.
6. Critically evaluate emerging risks involving data analytics, artificial intelligence and discrimination.

1. The Regulatory Purpose of Product Governance

Product governance is the system by which an insurer designs, approves, distributes, reviews and, where necessary, withdraws an insurance product. Its central question is not simply whether a product can be sold, but whether it is suitable, lawful, financially sustainable and consistent with the interests of its intended customers.

The principal regulatory objectives are:

• Solvency and policyholder protection: premiums and reserves must support the insurer’s obligations.
• Fair treatment: customers should receive products and services that correspond to their needs and expectations.
• Transparency: material benefits, limitations, conditions and costs must be disclosed.
• Actuarial soundness: pricing and reserving should be based on credible risk and cost information.
• Non-discrimination: distinctions must have a legitimate risk-related basis and comply with equality law.
• Market integrity: product design should not encourage mis-selling, hidden charges or regulatory arbitrage.

Product governance is both an ex ante and ex post activity. Ex ante controls operate before launch through product approval, legal review and actuarial certification. Ex post controls operate after launch through claims analysis, complaints, lapse monitoring, audit, customer outcomes testing and product remediation.

2. Governance Architecture and Accountability

Effective governance allocates responsibility among the board, senior management, actuarial function, risk function, compliance function, product owners, underwriting personnel and distributors. Responsibility cannot be transferred merely because a distributor or technology provider performs the operational task.

A robust product file normally contains:

The board should approve the governance policy and material products, while management implements controls. The actuarial function should independently assess assumptions, rating factors, premium adequacy and policyholder equity. Compliance should examine consumer law, licensing, disclosure, prohibited discrimination and sales practices. Internal audit should test the effectiveness of the entire framework rather than simply confirm that documents exist.

A useful governance principle is clear ownership: every material product should have a named owner responsible for its performance, customer outcomes, regulatory compliance and timely review.

3. Product Design and Approval

3.1 Needs-based design

Product development should begin with an identified insurable need, not with a sales target. The design team should define the risk event, the financial consequences of that event, the intended customer group and the level of protection sought. A product designed for mass-market consumers should not depend on sophisticated interpretation of technical conditions.

The target market should be described positively and negatively. For example, a household contents policy may be intended for individuals who occupy a specified residential property, but unsuitable for commercial premises, vacant properties or customers requiring unusually high-value specialist cover.

3.2 Approval stages

A typical approval process includes:

1. Concept approval: confirmation that the proposed product addresses a legitimate need and fits the insurer’s strategy.
2. Technical design: specification of benefits, eligibility, limits, deductibles, exclusions, premium basis and claims processes.
3. Actuarial assessment: analysis of frequency, severity, expenses, capital costs, uncertainty and profitability.
4. Legal review: examination of formation, interpretation, mandatory terms, cancellation rights, jurisdiction and compliance.
5. Conduct review: testing of suitability, disclosure, accessibility, vulnerability and distribution incentives.
6. Operational readiness: confirmation that systems, staff, claims procedures and complaint mechanisms can deliver the promise.
7. Formal approval and controlled launch: documented authorisation, version control and post-launch monitoring.

No product should be marketed before policy documents, rating tables, systems and training materials are aligned. A common governance failure occurs when a product is approved centrally but sold using outdated brochures or distributor-specific wording.

3.3 Material change control

A material change may include a new exclusion, revised eligibility rule, altered risk classification, significant premium change, new distribution channel, algorithmic pricing model or change in claims settlement practice. Material changes should receive approval proportionate to their effect. Minor administrative changes may use delegated approval, but changes affecting coverage or customer outcomes require renewed legal, actuarial and conduct assessment.

4. Underwriting and Risk Classification

Underwriting is the process of selecting, classifying and pricing risks within the insurer’s authority and appetite. It converts heterogeneous applicants into risk categories while applying contractual, legal and operational controls.

4.1 Underwriting principles

A sound underwriting system should be:

• Relevant: variables must relate to the insured risk.
• Reliable: information should be sufficiently accurate and consistent.
• Proportionate: questions and evidence requirements should not exceed the legitimate underwriting purpose.
• Consistent: similar risks should receive comparable treatment.
• Explainable: decisions should be capable of being communicated to customers and supervisors.
• Reviewable: declined, rated or referred cases should be subject to governance and appeal mechanisms.

Underwriting authority should be documented through limits, referral rules, prohibited risks, exception controls and audit trails. Delegated underwriting by agents, managing general agents or digital platforms does not remove the insurer’s responsibility for supervision.

4.2 Classification and discrimination

Risk classification is not automatically unfair merely because it produces different premiums. Insurance depends on differentiation. The legal issue is whether the distinction is permitted, risk-relevant, evidence-based and proportionate. Some factors may be restricted or prohibited even if statistically predictive, because equality law or public policy limits their use.

Examples requiring particular care include age, sex, disability, genetic information, race, religion, location, occupation and socioeconomic proxies. A variable can create indirect discrimination when it appears neutral but functions as a proxy for a protected characteristic. Governance should therefore examine both the variable and its interaction with other variables.

4.3 Information duties and misrepresentation

The legal consequences of inaccurate information depend on the applicable law and policy type. Some regimes use a consumer duty of reasonable care; others distinguish innocent, negligent and fraudulent misrepresentation. Product governance should make questions clear, avoid ambiguous forms and explain the consequences of non-disclosure. Insurers should not rely on obscure questions to create technical grounds for avoidance.

5. Pricing Regulation and Actuarial Fairness

Pricing must be adequate to meet claims and expenses, fair in its treatment of comparable risks, and compliant with regulatory restrictions. The fundamental pricing identity is:

where P is the premium, E[L] is expected claim cost, E[Ex] is expected operating expense, C is the cost of capital and risk margin, M is any applicable taxes or levies, and Π is the intended profit or contribution. The equation is a conceptual decomposition, not a universal tariff formula. It assumes that all relevant costs have been identified and that components are measured on a consistent basis.

For a simple risk pool, expected claim cost may be represented as:

where λ is expected claim frequency per exposure unit and s is expected claim severity in monetary units per claim. This multiplication is appropriate when frequency and severity are defined consistently and their dependence does not require a more complex model. If frequency and severity vary by policyholder or are dependent, the insurer must model the conditional distribution rather than apply a single pool average.

A technical premium may be expressed as:

where RM is a risk margin or capital-related allowance. A commercial premium may then include taxes, commissions and profit. Premium adequacy does not by itself establish fairness: a mathematically accurate price can still be unlawful if it uses prohibited factors or is presented deceptively.

Worked example: simplified risk premium

Suppose a portfolio has an expected frequency of 0.04 claims per policy-year and expected severity of ₹50,000 per claim. Expected claim cost is:

If expected expenses are ₹600, the risk margin is ₹200 and taxes or levies are ₹240, the premium before profit is:

This is a simplified illustration. It does not account for investment income, inflation, reinsurance, catastrophe accumulation, parameter uncertainty, policy duration or distribution commission. Those items must be incorporated where material.

5.1 Adequacy, fairness and affordability

Three principles may conflict. A premium must be adequate to avoid deterioration of solvency; it should reflect risk sufficiently to prevent systematic cross-subsidy that is not intended or disclosed; and it should remain accessible to the target market. Regulation may therefore permit or require social pooling, premium caps, community rating, subsidies or mandated coverage. In such cases, the cross-subsidy should be transparent and supported by the legal framework.

5.2 Governance of rating models

Rating models require documentation of data sources, variables, segmentation, assumptions, validation, limitations, overrides and monitoring. Model performance should be tested for calibration, stability, error rates and disparate outcomes. Changes to a model should be version-controlled, independently reviewed and approved before implementation.

6. Policy Documentation, Exclusions and Interpretation

The policy contract should clearly identify the parties, insured interest, period, premium, coverage trigger, limits, deductibles, conditions, exclusions, claims procedure, cancellation rights and dispute mechanisms. Documents should be internally consistent: the schedule, wording, endorsements, proposal form and marketing material must not contradict one another.

An exclusion defines the boundary of the insurer’s promise. Valid exclusions may address moral hazard, uninsurable events, accumulation, maintenance obligations, war, nuclear risks or risks intended for separate cover. However, exclusions should be prominent, intelligible and connected to the product’s purpose. A broad exclusion that defeats the reasonable central expectation of cover may attract regulatory or judicial scrutiny.

The contra proferentem principle, where recognised, may resolve genuine ambiguity against the party responsible for drafting. It is not a substitute for clear drafting. Insurers should use plain language, examples where useful, prominent presentation of significant limitations and translations or accessible formats required by law.

Claims conditions should not impose unreasonable procedural barriers. A notice requirement should distinguish between information necessary to investigate a claim and technical formalities that cause forfeiture without meaningful prejudice to the insurer.

7. Suitability, Distribution and Customer Outcomes

Suitability is especially important in life, health, investment-linked and complex commercial products. The distributor should understand the customer’s needs, resources, risk tolerance, existing cover and relevant vulnerability. A product that is appropriate for one customer may be unsuitable for another even if both meet formal eligibility criteria.

The insurer should control:

• sales scripts and illustrations;
• remuneration and incentive structures;
• replacement and switching practices;
• telemarketing and digital journeys;
• advice and non-advised sales distinctions;
• cooling-off and cancellation rights;
• treatment of vulnerable customers; and
• complaint and redress processes.

A disclosure document is not effective merely because it is technically complete. It should enable a reasonable customer to understand what is covered, what is not covered, how much is payable, what must be done after a loss and what circumstances may reduce or defeat a claim.

8. Application Across Major Insurance Classes

Life products may require long-term assumption governance and fairness between generations of policyholders. Health products require careful treatment of medical evidence and exclusions because contractual decisions can directly affect access to essential services. Property and catastrophe products require scenario analysis for correlated losses. Liability products require strong claims-made wording and reserve governance because the ultimate cost may remain uncertain for many years.

9. Lifecycle Monitoring and Remediation

A product is not compliant forever merely because it passed initial approval. Monitoring should combine quantitative and qualitative indicators, including:

• claims acceptance and repudiation rates;
• average settlement time;
• complaints by product, channel and demographic group;
• cancellations, lapses and renewal rates;
• premium changes and affordability indicators;
• loss ratios and reserve deterioration;
• override and exception rates;
• distribution remuneration;
• litigation and ombudsman outcomes; and
• model drift and data-quality failures.

A useful loss-ratio measure is:

where incurred claims generally include claims paid plus the change in outstanding claim liabilities, subject to the insurer’s accounting framework. Both numerator and denominator must relate to the same period and portfolio basis. The ratio is informative but incomplete: a low ratio may reflect delayed reporting, inadequate claims handling or insufficient coverage, while a high ratio may result from catastrophe events or deliberate social pricing.

Remedial action may include clarifying documents, retraining distributors, correcting premiums, paying affected customers, changing underwriting rules, withdrawing the product or notifying the supervisor. Remediation should address root causes rather than only individual complaints.

10. Technology, Artificial Intelligence and Data Governance

Digital underwriting and AI can improve speed, fraud detection and risk segmentation, but they introduce opacity, data-protection and discrimination risks. Governance should require lawful data collection, purpose limitation, data minimisation, security, explainability proportionate to impact, human escalation and effective correction rights.

An insurer should be able to answer: What data influenced the decision? Was the data accurate? Could the customer challenge the outcome? Does the model produce systematically adverse results for a protected or vulnerable group? Is the model still valid when economic, climatic or behavioural conditions change?

AI governance should include model inventory, impact assessment, independent validation, performance thresholds, incident reporting, supplier oversight and a documented prohibition on fully automated decisions where law or risk severity requires human review. Automation does not eliminate fiduciary, contractual or regulatory accountability.

11. Illustrative Case: Algorithmic Home Insurance Pricing

Background

An insurer introduced a digital home-insurance product using property location, construction information, claims history and external data to generate instant quotations.

Problem

After launch, complaints increased in several low-income neighbourhoods. The model did not directly use income or ethnicity, but location-based variables and property-data gaps produced materially higher premiums and more referral decisions in those areas. Customers also reported that the digital explanation—“risk outside appetite”—was inadequate.

Analysis

The product team had obtained actuarial approval but had not completed a conduct-impact assessment. Validation focused on aggregate predictive accuracy rather than outcome differences between customer groups. Data quality was poorer for older properties, and the model treated missing information as a high-risk signal. The policy wording was clear, but the quotation process did not explain the principal rating factors or provide a meaningful appeal route.

Approach and solution

The insurer paused expansion, conducted a fairness and data-quality review, replaced the missing-data rule with a neutral referral process, introduced human review for high-impact decisions, and required the model owner to document variables and limitations. It also created a customer explanation notice, reviewed historical quotes for possible remediation, and added monitoring of referral rates, premium distributions and complaints by relevant demographic and geographic categories.

Outcome

Predictive performance declined slightly, but unexplained referral differences narrowed. Customer complaints fell and the insurer obtained stronger evidence that the model was both technically useful and conduct-compliant.

Learning points

1. Statistical accuracy is not the same as legal or ethical acceptability.
2. Proxy discrimination can arise without explicit use of protected characteristics.
3. Missing data require governance, not automatic adverse assumptions.
4. Product approval must include customer-outcome testing.
5. Human review and redress are essential for high-impact automated decisions.

12. Challenges and Critical Evaluation

Product governance must manage tensions between innovation and precaution, risk-based pricing and solidarity, disclosure and comprehensibility, automation and accountability, and commercial speed and independent review. Excessive regulation may reduce product availability or encourage standardisation; weak regulation may permit opaque pricing, exclusionary design and solvency-threatening underpricing.

A further challenge is regulatory fragmentation. An insurer may face insurance supervision, consumer protection, data protection, competition, equality, tax and financial-crime obligations simultaneously. Governance should therefore use integrated compliance mapping rather than separate checklists.

Best practice is proportionate governance: simple, low-risk products may use streamlined review, while complex, long-term or high-impact products require deeper actuarial, legal, conduct and model scrutiny. Proportionality must reduce unnecessary process, not weaken substantive customer protection.

Chapter Summary

Product governance regulates the complete life of an insurance product, from needs identification and design through approval, distribution, monitoring and withdrawal. It requires coordinated accountability among the board, management, actuarial, risk, compliance, audit and distribution functions. Underwriting should use relevant, reliable and lawful risk factors, while pricing must be adequate, transparent and consistent with actuarial and equality principles. Policy documents must clearly express coverage, conditions and exclusions. Suitability and fair treatment require attention to customer needs, vulnerability, incentives and accessible disclosure. Lifecycle monitoring should use claims, complaints, lapse, pricing, model and fairness indicators. Digital and AI-based systems offer efficiency but require data governance, explainability, validation, human oversight and remediation. The central principle is that technical profitability or predictive accuracy cannot substitute for lawful, fair and sustainable customer outcomes.

Key Terms

Product governance: System for designing, approving, distributing, reviewing and withdrawing insurance products.

Target market: Customer group for whom a product is intended to be appropriate.

Underwriting: Selection, assessment and classification of insurance risks.

Risk classification: Grouping risks according to relevant characteristics for acceptance and pricing.

Actuarial fairness: Consistency between risk evidence, pricing treatment and applicable legal standards.

Premium adequacy: Sufficiency of premium to meet expected claims, expenses and risk costs.

Exclusion: Contractual provision removing specified risks or circumstances from coverage.

Suitability: Appropriateness of a product for a particular customer’s needs and circumstances.

Material change: Change capable of significantly affecting coverage, price, risk or customer outcomes.

Loss ratio: Incurred claims divided by earned premium.

Risk margin: Allowance for uncertainty, capital cost or adverse deviation.

Proxy discrimination: Disparate treatment produced by a variable that indirectly represents a protected characteristic.

Model governance: Controls over development, validation, use, monitoring and alteration of analytical models.

Vulnerability: Circumstance that may reduce a customer’s ability to understand, purchase or claim under insurance.

Lifecycle monitoring: Continuing review of product performance and customer outcomes after launch.

Delegated authority: Permission given to another party to underwrite or administer within defined limits.

Claims-made policy: Policy responding to claims made during the policy period, subject to its terms.

Parametric insurance: Cover triggered by a specified measurable parameter rather than assessed loss alone.

Remediation: Corrective action to repair customer, legal, operational or governance deficiencies.

Review Questions

Short Answer Questions

1. What are the principal objectives of insurance product governance?
2. Why must a product have a defined target market?
3. Distinguish premium adequacy from actuarial fairness.
4. What information should a product file contain?
5. Why is lifecycle monitoring necessary after product approval?

Long Answer Questions

1. Explain the stages of a robust insurance product approval process.
2. Discuss the legal and actuarial issues involved in risk classification.
3. Analyse the importance of clear policy wording and carefully drafted exclusions.
4. Explain how suitability and remuneration controls influence insurance distribution.
5. Evaluate the governance challenges created by AI-based underwriting and pricing.

Analytical / Application Questions

1. An insurer’s motor model reduces premiums for some customers but produces substantially higher prices for a group identified mainly through a location proxy. Design a legal, actuarial and conduct review.
2. A health insurer has a low loss ratio but unusually high claim repudiation and complaint rates. What conclusions can and cannot be drawn, and what monitoring should follow?
3. A property insurer launches catastrophe cover without testing accumulation exposure across neighbouring regions. Identify the governance failures and propose corrective measures.

Multiple Choice Questions with Answers

Question 1

Which activity is primarily an ex ante product-governance control?

A. Post-launch complaint analysis

B. Pre-launch actuarial and legal approval

C. Historical claims remediation

D. Annual lapse investigation

Correct Answer: B. Pre-launch actuarial and legal approval

Question 2

In the simplified relationship E[L] = λ × s, what does λ represent?

A. Expected claim severity

B. Expected operating expense

C. Expected claim frequency

D. Risk margin

Correct Answer: C. Expected claim frequency

Question 3

Which statement best describes actuarial fairness?

A. Every customer pays the same premium

B. Premiums are always as low as possible

C. Risk distinctions are evidence-based and legally permissible

D. Pricing is based only on customer income

Correct Answer: C. Risk distinctions are evidence-based and legally permissible

Question 4

What is a material product change?

A. Correction of a typographical error with no customer effect

B. A change to an exclusion affecting coverage

C. Reordering internal filing folders

D. Updating an employee’s job title

Correct Answer: B. A change to an exclusion affecting coverage

Question 5

A loss ratio is generally calculated as:

A. Earned premium divided by incurred claims

B. Incurred claims divided by earned premium

C. Expenses divided by capital

D. Premium tax divided by claims frequency

Correct Answer: B. Incurred claims divided by earned premium

Question 6

Why can a statistically predictive variable still be unacceptable?

A. All predictive variables are prohibited

B. It may create unlawful discrimination or use prohibited information

C. Predictive variables cannot improve underwriting

D. It necessarily reduces solvency

Correct Answer: B. It may create unlawful discrimination or use prohibited information

Question 7

Which is a suitable lifecycle-monitoring indicator?

A. Office electricity consumption only

B. Claims repudiation and complaint rates

C. Number of board meetings unrelated to the product

D. Competitor advertising colour

Correct Answer: B. Claims repudiation and complaint rates

Question 8

What is the principal function of an exclusion?

A. To define a risk or circumstance outside the coverage promise

B. To guarantee every possible loss

C. To replace premium collection

D. To eliminate the need for policy wording

Correct Answer: A. To define a risk or circumstance outside the coverage promise

Question 9

In the illustrative AI case, what was a major governance failure?

A. The model used no data at all

B. Validation considered predictive accuracy but not customer-outcome disparities

C. The insurer offered human review from the outset

D. The policy had no exclusions

Correct Answer: B. Validation considered predictive accuracy but not customer-outcome disparities

Question 10

Which principle is most relevant when an automated decision has a high impact on a customer?

A. No explanation is ever required

B. Human escalation and meaningful redress

C. Automatic rejection of all referrals

D. Use of the maximum available data regardless of purpose

Correct Answer: B. Human escalation and meaningful redress