Insurance is an institutional method for managing uncertain financial loss. Individuals, households, firms and public bodies face events such as death, illness, accidents, property damage, litigation and longevity. These events are uncertain for a particular policyholder, but their aggregate behaviour can often be estimated from historical and current data. Actuarial mathematics converts that uncertainty into financial values used for product design, pricing, reserving, capital management and strategic decision-making.

An actuary therefore does more than calculate premiums. The actuary evaluates assumptions, measures uncertainty, communicates financial consequences and helps ensure that promises made by an insurer remain sustainable. Actuarial work connects probability, statistics, economics, finance, accounting, regulation, data science and professional judgement.

Learning Objectives

1. The Economic Purpose of Insurance

Risk is the possibility that an uncertain event will produce an adverse consequence. Uncertainty may concern whether an event occurs, when it occurs, how severe it is, or how its consequences interact with other events. Insurance is concerned primarily with measurable financial risks, although legal, operational, reputational and strategic risks also affect insurers.

The insurer does not eliminate the underlying event. A life insurer cannot prevent death, and a property insurer cannot prevent every fire. It provides financial compensation or services when an insured event occurs.

Insurance is economically viable when the insurer can estimate the aggregate cost of claims, charge adequate premiums, control expenses, diversify risks and maintain sufficient capital. It is not necessary that every policyholder receive more in claims than they pay in premiums. The value of insurance lies in reducing the financial impact of adverse outcomes and converting uncertain large losses into more predictable payments.

2. Fundamental Actuarial Terminology

Term	Meaning
Exposure	A unit that is subject to possible loss, such as a vehicle-year or person-year.
Peril	The cause of loss, such as fire, theft or illness.
Hazard	A condition that increases the likelihood or severity of loss.
Claim	A demand for payment or benefit under an insurance contract.
Frequency	The number of claims relative to exposures.
Severity	The amount of loss per claim.
Premium	The amount charged for insurance cover.
Deductible	The portion of a covered loss borne by the policyholder.
Limit	The maximum amount payable under a policy or coverage.
Benefit	The contractual payment or service provided by the insurer.
Underwriting	Selection, classification and terms applied to risks.
Policy term	The period during which cover is in force.
Incurred claim	A claim arising from an insured event, whether reported or not.
Outstanding claim	An incurred claim not yet fully paid or settled.
Reserve	An amount held for future obligations and expenses.
Reinsurance	Insurance purchased by an insurer from another insurer.
Solvency	The ability to meet obligations as they fall due, including under adverse conditions.
Capital	Financial resources available to absorb unexpected losses.
Liability	A present or future obligation of the insurer.
Actuarial assumption	A quantified expectation about future experience, such as mortality or inflation.

3. The Mathematics of Risk Pooling

Let N be the number of exposures during a period and let X_i be the loss associated with exposure i. The aggregate loss is:

Here, S is measured in currency. If the number of exposures is fixed and losses are identically distributed with expected loss E[X] and variance Var(X), then:

These expressions assume independence and a fixed N. If losses are dependent, the variance includes covariance terms:

Positive dependence increases aggregate uncertainty and weakens the diversification benefit. This is important for catastrophes, pandemics, market-linked guarantees and geographically concentrated portfolios.

For a portfolio with independent risks, the standard deviation of aggregate loss grows approximately with √N, whereas expected loss grows with N. Consequently, the coefficient of variation decreases as the portfolio becomes larger. This is the mathematical basis of pooling, but it is not a guarantee of profit. Parameter uncertainty, model error, fraud, inflation, changing behaviour and extreme events remain.

where K is the random number of claims and Y_j is the severity of claim j. If K is independent of the severities, with mean E[K] and variance Var(K), then:

This decomposition separates uncertainty caused by the number of claims from uncertainty caused by claim amounts. It is widely used in general insurance pricing and capital modelling.

A simple pricing illustration

Suppose a portfolio has 10,000 vehicle-years. The expected claim frequency is 0.08 claims per vehicle-year and the expected claim severity is 2,500 currency units. Expected claims are therefore:

If expenses are 400,000 and the insurer requires 300,000 for risk margin and profit, the indicated total premium is 2,700,000, or 270 per vehicle-year before taxes and any adjustments. This is an expected-value illustration, not a complete commercial tariff. A real tariff would consider credibility, expenses by policy, reinsurance, investment income, capital costs, competition, trends, limits, deductibles and regulatory requirements.

4. Insurance Contracts and Product Classes

Life insurance

Life products depend on survival and death probabilities, timing of benefits, expenses, investment returns and policyholder behaviour. Major forms include:

Health and disability insurance

Health insurance covers medical costs or services. Disability insurance may provide income replacement when illness or injury prevents work. Important actuarial issues include incidence, duration, recovery, medical inflation, provider behaviour, benefit coordination and moral hazard.

Property and casualty insurance

Property insurance covers physical assets and associated financial loss. Casualty insurance includes motor, liability, workers’ compensation, marine and other non-life classes. These products commonly involve annual contracts, deductibles, limits, claims development and substantial inflation or legal uncertainty.

Reinsurance

Under proportional reinsurance, the reinsurer receives an agreed share of premiums and pays the same share of claims. Under excess-of-loss reinsurance, the reinsurer pays losses above a specified attachment point up to a limit. Reinsurance can reduce volatility, protect capital and increase underwriting capacity, but it introduces counterparty, basis and liquidity risks.

5. The Actuarial Control Cycle

The actuarial control cycle is a continuous framework for decision-making under uncertainty. It is not a one-time calculation. The cycle normally contains the following stages:

The cycle links technical analysis with governance. A sophisticated model that is poorly implemented or not monitored can produce worse decisions than a simpler, transparent model.

6. Actuarial Mathematics in Core Decisions

Pricing

Pricing estimates the present value of expected benefits and expenses and adds allowances for risk, capital, profit and other costs. A simplified gross premium relationship is:

The exact formulation depends on timing, premium frequency, taxes, commissions, investment income and accounting or regulatory rules. Premiums must also satisfy contractual, fairness and market constraints. A technically adequate premium may be commercially unsuitable, while a low premium may create future solvency problems.

Reserving

A reserve is a liability estimate for future payments and expenses relating to obligations already incurred or contracts already in force. Non-life reserves commonly include:

where C_t is the future cash flow at time t, T is the final projection period, and v = 1/(1+i) is the discount factor when the annual effective discount rate is i. Discounting is appropriate only when consistent with the applicable accounting, regulatory and actuarial framework and when the timing and risk adjustment have been properly considered.

Life reserves may be calculated prospectively as the present value of future benefits and expenses minus the present value of future premiums, under specified assumptions. A reserve is not a fund in a separate bank account; it is an estimate of a liability supported by assets and capital.

Capital and solvency

Expected loss is not sufficient for solvency management. Capital protects against adverse deviations from expected outcomes. A simplified economic capital measure may be represented as:

where α is a selected confidence level and the loss distribution, horizon, diversification and risk aggregation method must be specified. This is a conceptual expression, not a universal regulatory formula. Regulatory capital frameworks may use prescribed stresses, standard formulas, internal models or combinations of these approaches.

Capital decisions consider underwriting, market, credit, operational, liquidity and concentration risks. The insurer must also assess capital quality, fungibility, group structure, stress scenarios and recovery options.

Management information

Actuarial analysis supports underwriting authority, product redesign, claims strategy, reinsurance purchase, investment policy, distribution decisions and performance measurement. Useful indicators include loss ratios, expense ratios, combined ratios, persistency, mortality experience, claim settlement speed and capital consumption. Each metric requires careful definition; for example, a loss ratio based on paid claims differs from one based on incurred claims.

7. Stakeholders and Professional Responsibilities

The policyholder pays premiums and supplies information, expecting valid claims to be paid fairly. The insurer designs contracts, underwrites risks, invests assets, manages claims and maintains solvency. Intermediaries advise customers, distribute products and may assist with claims. Reinsurers provide risk capacity and diversification. Regulators and supervisors protect policyholders, promote market integrity and monitor solvency and conduct. Investors and boards provide capital and oversee strategy. Employees, claims professionals and technology providers influence operational outcomes.

The actuary has a responsibility to provide competent, objective and appropriately qualified advice. Important professional principles include integrity, confidentiality, competence, transparency about uncertainty, documentation, independence of judgement and escalation of material concerns. Ethical actuarial practice requires consideration of customers who may be vulnerable, unfair discrimination, accessibility, privacy and the consequences of automated decisions.

8. Data, Models and Contemporary Developments

Insurance increasingly uses telematics, electronic health records, geospatial data, connected devices and machine-learning methods. These tools may improve risk classification and claims handling, but they do not remove actuarial responsibility. Key controls include data lineage, validation, explainability, cybersecurity, bias testing, human oversight and monitoring for changes in the underlying population.

Climate change creates non-stationary hazards, particularly in property and agricultural insurance. Historical averages may no longer be reliable guides to future catastrophe frequency or severity. Pandemic risk, cyber risk, longevity improvement, inflation volatility, supply-chain disruption and geopolitical uncertainty similarly require scenario analysis rather than reliance on a single central estimate.

Sensitivity testing, stress testing and reverse stress testing help decision-makers understand vulnerability. Model outputs should be accompanied by assumptions, limitations, ranges and management actions.

9. Illustrative Case: Motor Portfolio Deterioration

Background

An insurer writes 50,000 private motor vehicle-years. At the start of the year, the pricing model assumed a frequency of 0.10 claims per vehicle-year and an average severity of 3,000 currency units. The indicated expected claims were:

Problem

After nine months, reported claims were materially above plan. Repair inflation had increased, driving patterns had changed, and a large urban segment had higher theft frequency. Management initially proposed a uniform premium increase.

Analysis

The actuary separated frequency and severity experience by territory, vehicle age, coverage and distribution channel. The investigation found that frequency was especially high in two territories, while severity had increased across most territories because of parts and labour inflation. Claims development analysis also indicated that early reported data understated ultimate costs.

The actuary tested several actions: revised territorial factors, anti-theft requirements, claims fraud screening, repair-network negotiation, a revised deductible and targeted reinsurance. A uniform increase would have overcharged lower-risk customers and might have encouraged profitable low-risk policyholders to leave.

Approach and outcome

The insurer adopted segmented underwriting, updated severity trends, strengthened claims controls and introduced a monitored deductible option. The revised model was subjected to independent validation and documented for regulatory review. Management received a range of projected outcomes rather than a single forecast. The result was improved risk selection and a more defensible premium structure, although uncertainty remained because future inflation and theft patterns were not known with certainty.

Learning points

The case demonstrates that actuarial control is a cycle rather than a once-yearly tariff exercise. It also shows why aggregate results must be decomposed into frequency, severity, mix, inflation and claims development. Fairness, customer communication, model governance and commercial consequences are part of actuarial decision-making.

Challenges and Limitations

Actuarial results can be misleading when data are incomplete, biased, inconsistent or not representative of future business. Historical data may reflect old underwriting rules or suppressed demand. Small portfolios may lack credibility. Rare catastrophes provide little direct experience. Correlation may change under stress, and assumptions that are reasonable individually may be inconsistent collectively.

Other limitations include parameter uncertainty, changing regulation, behavioural response to prices, moral hazard, adverse selection and conflicts between accounting, economic and regulatory measures. Professional judgement is unavoidable, but it should be explicit, challenged, documented and reviewed. No model can replace governance, expert scrutiny and prudent management.

Best-Practice Principles

Chapter Summary

Actuarial mathematics applies probability, statistics, finance and economics to uncertain insurance cash flows. Insurance works through risk transfer and pooling, but diversification is limited by dependence, concentration and extreme events. Frequency–severity models explain aggregate claims, while assumptions about mortality, morbidity, lapse, inflation, expenses and investment returns support pricing and reserving.

The actuarial control cycle links problem definition, data analysis, assumption setting, modelling, implementation, monitoring and revision. Actuaries support premiums, reserves, capital, reinsurance and management information. Their work must account for stakeholders, regulation, ethics, model risk and customer outcomes. Modern developments such as artificial intelligence and climate-risk modelling increase analytical capability but also require stronger governance and transparency.

Key Terms

Actuarial control cycle: Iterative framework for actuarial analysis and decision-making.

Moral hazard: Change in behaviour because insurance reduces financial consequences.

Review Questions

Short Answer Questions

Long Answer Questions

Analytical / Application Questions

Multiple Choice Questions with Answers

Question 1

B. Many policyholders contribute to a common fund from which covered losses are paid

Correct Answer: B. Many policyholders contribute to a common fund from which covered losses are paid

Question 2

If claim frequency is 0.08 and expected severity is 2,500 currency units, expected loss per exposure is:

Question 3

Correct Answer: B. Setting assumptions and subsequently monitoring experience

Question 4

Correct Answer: B. The policyholder retains an initial portion of a covered loss

Question 5

Under the independence assumptions for a fixed number of identical risks, aggregate expected loss equals:

Correct Answer: A. The number of exposures multiplied by expected loss per exposure

Question 6

Which reserve most directly relates to claims that have occurred but have not yet been reported?

Question 7

Correct Answer: C. It can increase aggregate loss volatility and capital requirements

Question 8

B. Uncertainty about the true future claim frequency estimated from limited data

Correct Answer: B. Uncertainty about the true future claim frequency estimated from limited data

Question 9

Question 10

Which practice best supports responsible use of artificial intelligence in insurance?

Correct Answer: C. Monitoring bias, explainability, data quality and human oversight

Chapter 1 Actuarial Mathematics and the Insurance Risk Framework

Actuarial Mathematics and the Insurance Risk Framework

Introduction