Introduction to Actuarial Mathematics

Chapter 10
Pension, Annuity, Health and Long-Term Benefit Applications

Post Graduate Program in Insurance Management
Introduction to Actuarial Mathematics   •   Post Graduate

Pension, Annuity, Health and Long-Term Benefit Applications

Introduction and context

Actuarial mathematics becomes most consequential when it is applied to benefits that are paid over uncertain periods and are influenced by mortality, morbidity, longevity, inflation, employment, investment returns and human behaviour. Pensions, life annuities, disability income, health insurance and employee benefits all require the actuary to convert uncertain future cash flows into present values, premiums, reserves and funding contributions.

These applications share a common structure: a contingent benefit is paid if a specified event occurs, the timing of payment is uncertain, and the financial value of the payment depends on discounting. However, each application introduces distinctive risks. A pension plan is exposed to longevity and salary risk; a health plan is exposed to morbidity, medical inflation and utilisation; a disability policy is exposed to incidence, recovery and claim termination; and an employee-benefit valuation is exposed to workforce turnover and compensation growth.

The actuarial task is therefore not merely to calculate a single expected value. It is to construct a defensible model, select assumptions, test sensitivity, allocate costs fairly, and communicate uncertainty to trustees, management, regulators and members.

Learning objectives

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

  1. Value pension benefits and life annuities using survival probabilities and discount factors.
  2. Distinguish defined-benefit, defined-contribution and hybrid retirement arrangements.
  3. Explain longevity risk and evaluate basic funding responses.
  4. Model disability income and health-insurance cash flows using incidence, continuation and recovery assumptions.
  5. Apply actuarial present-value methods to employee benefits.
  6. Interpret the effects of inflation, salary growth, medical trends, expenses and investment returns.
  7. Evaluate product sustainability, governance, ethics and model risk.

1. The actuarial present-value framework

Let vt = (1+i)-t denote the discount factor for one unit payable at time t, where i is the effective annual interest rate and t is measured in years. If a benefit of amount B_t is payable at time t only when event E_t occurs, its actuarial present value is:

APV = ∑t=1n Bt · Pr(Et) · vt

Here, n is the maximum payment duration, Pr(E_t) is the probability that the payment condition is satisfied, and all monetary amounts must be expressed in the same currency and time basis. The formula assumes that the benefit amount and event probability are appropriately specified and that the interest rate represents the valuation basis.

For a continuous payment stream, the corresponding expression is:

APV = ∫₀T B(t) · Pr(Et) · e-δt dt

where B(t) is the payment rate at time t, T is the maximum duration and δ = ln(1+i) is the continuously compounded force of interest. The discrete and continuous forms should not be mixed without adjusting the timing convention.

The equivalence principle sets a premium or contribution equal to the expected present value of benefits and expenses, subject to the chosen risk margin:

PV(premiums) = PV(benefits) + PV(expenses) + risk margin

For a fully funded deterministic illustration, the risk margin may be zero. In commercial insurance, however, capital, adverse-deviation margins, profit requirements and regulatory constraints normally mean that the premium exceeds the best-estimate expected cost.

Survival and mortality notation

For a life aged x, let:

  • lx = number of persons alive at exact age x in a life table;
  • dx = lx - lx+1 = deaths between ages x and x+1;
  • px = lx+1/lx = one-year survival probability;
  • qx = dx/lx = 1-px = one-year mortality probability;
  • {}tpx = probability that a person aged x survives t years;
  • vt = discount factor for a payment at time t.

Under a life-table model, {}tpx = lx+t/lx when integer ages and the relevant table assumptions apply. The probability of death between times t and t+1 is {}tpx · qx+t.

2. Life annuities and retirement income

A life annuity pays while the annuitant remains alive. It converts accumulated retirement wealth into a stream of income and transfers longevity risk from the individual to the annuity provider or pension plan.

A whole-life annuity-immediate paying 1 at the end of each year while a life aged x is alive has actuarial present value:

äx = ∑k=1∞ vk · {}kpx

The notation äx is commonly used for an annuity-immediate in some actuarial conventions, although notation varies by textbook. A temporary annuity for n years is:

äx:n = ∑k=1n vk · {}kpx

A life annuity-due pays at the beginning of each year and is valued as:

ä̈x = ∑k=0∞ vk · {}kpx = 1 + ∑k=1∞ vk · {}kpx

The first payment is certain at time zero, assuming the annuity is purchased by a living person. An annuity payable monthly requires a monthly discount rate and payment-timing assumptions. A common approximation is to divide the annual payment by 12, but the provider must specify whether payments are monthly in advance or in arrears and how mortality within the year is handled.

Worked example: deferred annuity

A person aged 60 purchases an annuity paying 12,000 per year at the end of each year from age 65, provided the person is alive. Suppose the annual effective interest rate is 4%, and the assumed survival probabilities are:

Payment ageYears from purchaseSurvival probabilityPayment
6550.9012,000
6660.8712,000
6770.8312,000

The truncated value is:

APV = 12,000[(0.90)(1.04)-5 + (0.87)(1.04)-6 + (0.83)(1.04)-7]

Using approximate discount factors, v5 ≈ 0.8219, v6 ≈ 0.7903 and v7 ≈ 0.7599:

APV ≈ 12,000[0.90(0.8219) + 0.87(0.7903) + 0.83(0.7599)]
APV ≈ 12,000(2.096) ≈ 25,152

Thus, under the three-payment approximation, the present value is approximately 25,152 monetary units. A complete valuation would include all later payments, expenses, profit margins, options and any guarantee period.

Important annuity design features

  • Immediate versus deferred: a deferred annuity begins after a waiting period and is exposed to survival during deferral.
  • Life-only versus guaranteed: a life-only annuity stops at death; a guaranteed-period annuity continues to a beneficiary for the remaining guarantee period.
  • Level versus escalating: escalating benefits may preserve purchasing power but require a larger initial premium.
  • Single-life versus joint-life: joint-life benefits continue while one or both members survive, according to the contract definition.
  • Fixed versus variable: variable benefits may be linked to investment performance, introducing investment risk to the retiree.

3. Pension schemes and funding

Defined-contribution arrangements

In a defined-contribution plan, the contribution is specified but the eventual pension is uncertain. The retirement fund depends on contributions, investment returns, charges and the timing of retirement withdrawals. A simplified accumulation equation is:

Fn = F0(1+r)n + ∑t=1n Ct(1+r)n-t

where F_n is the fund at retirement, F_0 is the initial fund, r is the annual net investment return, C_t is the contribution at time t and n is the number of years. This deterministic formula assumes a constant return and known contributions; actual returns are stochastic and may be correlated with inflation and wages.

If the desired annual retirement income is R and the annuity factor at retirement is äy, the required fund is approximately:

Frequired = R · äy

where y is the retirement age. This expression excludes taxes, expenses, benefit escalation and capital requirements unless they are incorporated into the annuity factor.

Defined-benefit arrangements

In a defined-benefit plan, the pension is specified by a formula, often involving final salary or career-average salary, pensionable service and an accrual rate. A typical benefit is:

Annual pension = accrual rate × years of service × pensionable salary

For example, with an accrual rate of 1/60, 30 years of service and pensionable salary of 900,000, the annual pension is:

(1/60) × 30 × 900,000 = 450,000

The pension payable at retirement is then converted into a liability at the valuation date by allowing for future service, salary increases, retirement probabilities, mortality, withdrawal and discounting.

The projected-unit-credit approach generally attributes an additional portion of the projected ultimate benefit to each period of service. A simplified liability is:

Liability = ∑t [benefit attributed to service × probability of payment × discount factor]

The actual valuation must follow the applicable accounting, actuarial and regulatory framework. The discount rate may be based on high-quality corporate bond yields, government yields or an approved asset-based approach, depending on the governing rules.

Funding status and contribution requirements

A plan’s funding ratio is commonly expressed as:

Funding ratio = market value of assets ÷ actuarial liability

A ratio below 100% does not automatically imply insolvency, because liability definitions, discount rates, recovery plans and sponsor strength matter. The actuarial contribution may include:

  1. the normal cost for benefits accruing during the current period;
  2. expenses;
  3. a payment toward any deficit;
  4. a margin for adverse experience or risk management.

The discount rate has a substantial effect. A lower rate generally increases the present value of long-duration pension liabilities, while a higher rate reduces it, all else equal. This creates asset–liability management challenges because assets and liabilities may respond differently to interest-rate movements.

4. Longevity risk and sustainability

Longevity risk is the risk that people live longer than anticipated, increasing the duration and cost of pensions and annuities. It differs from ordinary mortality fluctuation: a large population may experience relatively stable random mortality but still face systematic improvement in life expectancy.

Longevity assumptions should distinguish:

  • current mortality level;
  • future mortality improvement;
  • cohort effects;
  • socioeconomic and geographic differences;
  • selection effects among annuitants;
  • uncertainty in medical technology and public health.

Useful risk-management techniques include conservative mortality assumptions, regular assumption reviews, benefit design changes, longevity swaps, reinsurance, buy-ins or buy-outs, and diversified asset–liability strategies. A robust analysis includes scenario testing, such as one-year increases in life expectancy, accelerated mortality improvement and adverse selection at annuity purchase.

Sustainability is not equivalent to affordability at inception. A product is sustainable only when expected premiums, reserves, capital and investment strategy remain adequate under plausible adverse experience. Management should monitor actual-to-expected mortality, persistency, expenses, inflation, benefit take-up and investment performance.

5. Disability income insurance

Disability income insurance pays an income when illness or injury prevents work. Its valuation requires more than mortality assumptions. Relevant states may include active employment, disabled and receiving benefits, disabled but in a waiting period, recovered, retired or dead.

A simplified annual claim-cost formula is:

Expected claim cost = exposure × incidence rate × average benefit duration × benefit amount

This is useful for preliminary pricing but is not sufficient for a mature valuation because incidence and termination vary by age, occupation, severity and duration since disability.

A multi-state model uses transition intensities. If μAD(t) is the intensity of transition from active to disabled, μDR(t) is the recovery intensity and μDD(t) is the mortality intensity while disabled, the probability of being in each state is obtained from transition equations. In matrix form:

p'(t) = p(t)Q(t)

where p(t) is a row vector of state probabilities and Q(t) is the transition-intensity matrix. Off-diagonal entries are transition intensities and each diagonal entry is the negative sum of the other entries in its row. The model is valid only when the state definitions, transition assumptions and time scale are consistent.

Benefits may be subject to an elimination period, benefit escalation, offsets for social insurance, partial-disability provisions and maximum benefit periods. Pricing must allow for claim incidence, duration, recovery, mortality, expenses, interest and behavioural responses.

6. Health insurance applications

Health insurance differs from life insurance because claims may recur, frequencies vary significantly and medical costs are affected by technology, provider prices and policyholder behaviour. A basic aggregate-claims model is:

S = ∑j=1N Xj

where S is aggregate claims, N is the random number of claims and X_j is the cost of claim j. If claim frequency and severity are independent, then:

E[S] = E[N] · E[X]

and

Var(S) = E[N]Var(X) + Var(N)[E(X)]²

These identities require finite moments and independence between N and claim severities. They are useful for pricing and capital analysis but may be inappropriate where large claims, provider networks or epidemic events create dependence.

Premium adequacy can be represented as:

Premium = expected claims + expenses + cost of capital + risk margin − expected recoveries

Expected claims should reflect age, plan design, deductibles, co-payments, limits, provider prices, utilisation, morbidity and medical trend. Medical trend may be decomposed into price inflation, service-intensity growth, technology effects and changes in population health.

A loss ratio is:

Loss ratio = incurred claims ÷ earned premium

It is an experience indicator, not a complete measure of profitability. Expenses, reserves, capital, reinsurance and investment income must also be considered.

7. Employee benefits and post-employment obligations

Employee benefits include pensions, post-retirement medical care, severance, long-service awards, disability benefits and leave obligations. Valuation requires demographic and financial assumptions and a clear attribution method.

For a benefit payable on a future event, a general model is:

APV = ∑t Benefitt × Pr(employed and eligible at t) × Pr(event at t | eligible) × vt

Employment probability may be expressed through turnover or withdrawal rates. Salary-linked benefits additionally require salary escalation. If salary grows at annual rate g, a current salary S_0 projected t years is:

St = S0(1+g)t

This is a deterministic assumption, not a forecast guarantee. For post-retirement medical benefits, the model must include retirement eligibility, participation, dependent coverage, medical inflation, claims utilisation and mortality.

The interaction between assumptions is important. Higher salary growth increases salary-related benefits; higher turnover may reduce long-service benefits but can increase immediate termination payments; higher medical inflation increases post-retirement medical liabilities; and a lower discount rate increases the present value of long-duration obligations.

8. Illustrative integrated case: Meridian Manufacturing

Background

Meridian Manufacturing, a fictional employer, sponsors a defined-benefit pension plan and provides post-retirement medical cover. The pension formula is 1/60 of final pensionable salary for each year of service. The plan has 800 active members and 240 pensioners.

Problem

The trustees observe a funding ratio of 92%. Pensioners are living longer than the mortality basis used five years ago, medical costs are increasing faster than general inflation, and the asset portfolio is concentrated in growth assets. Management wants to reduce contributions without weakening member security.

Analysis

The actuary performs a best-estimate valuation and stress tests three risks:

StressMain effect
Lower discount rateIncreases pension and medical liabilities
One-year longevity improvementIncreases pension duration and annuity cost
Medical trend +2 percentage pointsMaterially increases medical liabilities

The analysis separates pension liabilities from medical liabilities because their drivers differ. Pension liabilities are strongly exposed to interest rates and longevity. Medical liabilities are additionally exposed to healthcare utilisation, provider prices and plan design.

The actuary also compares the duration of liabilities with the duration of assets. A high allocation to equities may improve long-term expected return but creates contribution volatility and does not provide a close hedge against interest-rate changes.

Approach and solution

The trustees adopt a staged plan:

  1. update mortality and medical-cost assumptions using credible experience and external evidence;
  2. introduce a liability-driven investment allocation for a portion of pension liabilities;
  3. establish a deficit-recovery contribution schedule;
  4. introduce co-payments and managed-care controls for future medical coverage while protecting vulnerable members;
  5. monitor quarterly experience against expected claims, mortality and expenses;
  6. communicate clearly that assumption changes are not the same as benefit reductions.

Outcome

The plan does not eliminate risk, but it improves transparency and reduces the probability that a single adverse experience will cause an unexpected funding crisis. The trustees retain a formal review trigger if the funding ratio falls below an agreed level or if longevity and medical trends diverge materially from assumptions.

Learning points

The case demonstrates that actuarial sustainability requires integrated management of liabilities, assets, assumptions, governance and member communication. Reducing the valuation liability through an aggressive discount rate would not solve the underlying economic risk. Similarly, reducing medical benefits without analysing utilisation and equity consequences may transfer costs rather than improve sustainability.

9. Professional, ethical and regulatory considerations

Actuaries exercise judgement where evidence is incomplete. Important governance practices include documenting assumptions, explaining data limitations, conducting independent review, maintaining version control and distinguishing best estimate from prudence margins.

Fairness is central. Age, health status, occupation, income and family structure may be legitimate actuarial variables, but their use must comply with insurance, employment, privacy and anti-discrimination law. Risk classification can improve pricing accuracy while also creating affordability and access concerns.

Member communications should explain uncertainty, not imply guarantees that the model cannot support. Trustees must understand conflicts of interest, particularly where an adviser is also linked to an investment manager, insurer or benefit provider.

Regulatory regimes differ, but commonly require adequate technical provisions, capital, governance, data quality, stress testing and reporting. The actuary must apply the relevant local standards rather than assume that a formula valid in one jurisdiction is automatically acceptable in another.

10. Technology, data and contemporary developments

Modern valuations use administrative databases, electronic health records, claims platforms and predictive models. Machine-learning techniques can assist in forecasting claims, mortality improvement or lapse behaviour, but they do not remove actuarial responsibility. Models require validation, explainability, monitoring for drift, privacy protection and controls against discriminatory outcomes.

Climate change may affect health, mortality, disability and investment returns. Pandemic risk illustrates the importance of dependence between morbidity, mortality, employment and financial markets. Alternative retirement products, collective defined-contribution schemes and partial annuitisation seek to balance income security with flexibility, but they must be evaluated for governance, communication and intergenerational fairness.

Best practice is to combine statistical sophistication with actuarial judgement. A complex model with poor data, unclear assumptions or weak governance is less reliable than a transparent model whose limitations are understood.

Challenges and limitations

Actuarial applications face several persistent limitations:

  • Parameter uncertainty: mortality, morbidity, inflation and investment assumptions are estimated rather than known.
  • Model risk: a model may omit dependence, behavioural response or structural change.
  • Data limitations: claims may be incomplete, inconsistent or unrepresentative of future members.
  • Long-term extrapolation: pension and annuity liabilities extend beyond reliable historical experience.
  • Basis risk: assets used for hedging may not respond exactly like liabilities.
  • Selection and behaviour: healthier individuals may buy annuities, while policyholders may lapse selectively.
  • Regulatory change: accounting and solvency rules can change reported liabilities without changing economic cash flows.

Sensitivity analysis, scenario testing, stochastic simulation, independent peer review and regular experience monitoring are therefore essential complements to point estimates.

Chapter Summary

Actuarial applications in pensions, annuities, disability, health and employee benefits use a common present-value framework but require different demographic, financial and behavioural assumptions. Life annuities depend principally on survival, discounting, benefit design and longevity improvement. Pension valuations distinguish defined-contribution accumulation from defined-benefit liability and funding analysis. Disability models require transitions among active, disabled, recovered and deceased states. Health-insurance models combine claim frequency, severity, medical trend, expenses and risk margins. Employee-benefit valuations incorporate employment probability, eligibility, salary growth and event probabilities.

Sustainability depends on more than expected value. It requires adequate reserves, capital, asset–liability management, realistic assumptions, stress testing, governance and ethical treatment of members and policyholders. Technology can improve prediction, but model transparency, data quality and professional judgement remain indispensable.

Key Terms

Actuarial present value: Present value of expected contingent future payments.

Annuity-immediate: Annuity paid at the end of each payment period.

Annuity-due: Annuity paid at the beginning of each payment period.

Defined-benefit plan: Pension plan specifying a benefit formula.

Defined-contribution plan: Plan specifying contributions while retirement benefits depend on accumulated assets.

Discount factor: Present value of one monetary unit payable in the future.

Funding ratio: Plan assets divided by actuarial liabilities.

Longevity risk: Risk that people live longer than assumed.

Mortality rate: Probability or intensity of death over a specified period.

Survival probability: Probability that a person remains alive to a specified future time.

Disability incidence: Rate at which active persons become disabled.

Recovery rate: Rate at which disabled persons return to work or leave disability status.

Medical trend: Growth in healthcare costs arising from prices, utilisation and treatment changes.

Claim frequency: Number of claims relative to exposure.

Claim severity: Monetary cost of an individual claim.

Technical provision: Liability held for expected insurance obligations under a regulatory basis.

Normal cost: Cost attributed to benefits accruing during the current service period.

Adverse selection: Disproportionate participation by higher-risk individuals.

Asset–liability management: Coordinated management of assets and liability cash flows.

Risk margin: Amount added to reflect uncertainty, capital cost or adverse deviation.

Review Questions

Short Answer Questions

  1. What is the difference between an annuity-immediate and an annuity-due?
  2. Define longevity risk and explain why it is important for pension plans.
  3. State the principal difference between defined-benefit and defined-contribution plans.
  4. What do claim frequency and claim severity measure in health insurance?
  5. Why should salary growth be included in the valuation of salary-linked employee benefits?

Long Answer Questions

  1. Explain the actuarial present-value framework and discuss the assumptions required for a reliable valuation.
  2. Describe the main risks affecting a life-annuity provider and evaluate possible risk-management responses.
  3. Explain how a defined-benefit pension liability and funding ratio are determined.
  4. Discuss the construction and limitations of a multi-state disability-income model.
  5. Analyse the financial, ethical and governance issues involved in valuing post-retirement medical benefits.

Analytical / Application Questions

  1. A retiree aged 65 requires annual income of 300,000. Explain how the required retirement fund would be estimated using an annuity factor, and identify the assumptions that could materially change the result.
  2. A health insurer observes rising claim frequency but stable claim severity. Analyse the likely effect on premium adequacy and identify operational data that should be investigated.
  3. A pension scheme has a funding ratio of 88%, a high equity allocation and rapidly improving longevity. Recommend an integrated actuarial and investment response, including the risks of your recommendation.

Multiple Choice Questions with Answers

Question 1

Which expression best represents the actuarial present value of a discrete contingent benefit?

A. Benefit multiplied by probability of payment and discount factor, summed over payment dates

B. Benefit divided by the mortality rate only

C. Premium multiplied by the number of policyholders

D. Asset value divided by annual expenses

Correct Answer: A. Benefit multiplied by probability of payment and discount factor, summed over payment dates

Question 2

An annuity-due differs from an annuity-immediate because it pays:

A. Only if the annuitant dies

B. At the beginning rather than the end of each payment period

C. Only during a deferred period

D. At a variable interest rate

Correct Answer: B. At the beginning rather than the end of each payment period

Question 3

A defined-contribution pension plan primarily specifies:

A. The final pension payable

B. The mortality table to be used

C. The contributions paid into the plan

D. The guaranteed annuity factor at retirement

Correct Answer: C. The contributions paid into the plan

Question 4

If annual salary is 900,000, service is 30 years and the accrual rate is 1/60, the annual pension is:

A. 15,000

B. 30,000

C. 450,000

D. 1,800,000

Correct Answer: C. 450,000

Question 5

Longevity risk is the risk that:

A. Interest rates remain unchanged

B. People live longer than assumed

C. Claims are submitted electronically

D. Medical expenses fall unexpectedly

Correct Answer: B. People live longer than assumed

Question 6

In a basic aggregate-claims model, S = ∑X_j represents:

A. The total of individual claim amounts

B. The average policy premium

C. The number of policyholders

D. The discount rate

Correct Answer: A. The total of individual claim amounts

Question 7

Under independence of claim frequency and severity, expected aggregate claims equal:

A. Expected frequency plus expected severity

B. Expected frequency divided by expected severity

C. Expected frequency multiplied by expected severity

D. Variance of frequency multiplied by severity

Correct Answer: C. Expected frequency multiplied by expected severity

Question 8

A pension funding ratio is generally calculated as:

A. Liabilities divided by expenses

B. Market value of assets divided by actuarial liabilities

C. Premiums divided by claims

D. Salary divided by service

Correct Answer: B. Market value of assets divided by actuarial liabilities

Question 9

Which factor is especially important in post-retirement medical-benefit valuation?

A. Medical inflation and healthcare utilisation

B. Only the employee’s current salary

C. The insurer’s brand name

D. The number of investment analysts

Correct Answer: A. Medical inflation and healthcare utilisation

Question 10

The equation p′(t) = p(t)Q(t) is associated with:

A. A multi-state transition model

B. A simple compound-interest calculation

C. A pension accrual formula only

D. A claims loss-ratio calculation

Correct Answer: A. A multi-state transition model

Introduction to Actuarial Mathematics
Chapter 10 of 14