This guide develops the calculations and modeling judgments behind survival probabilities, state-dependent premiums and reserves, joint-life benefits, profit testing, pensions, universal life and embedded guarantees. Twenty original constructed-response exercises include complete data, model answers and assessable rubrics; sixty distinct concepts explain mechanisms, worked applications and common errors. The supplied October 2026 syllabus establishes relevance and format, while the technical teaching uses established actuarial methods with explicit assumptions. This representative preview is not a full mock: the examination has six questions, 60 points and three hours, including one Excel question. Coverage follows the supplied allocation, current as of October 17, 2026; subsequent versions are outside this guide.
What this guide covers
Prepare for October 2026 ALTAM with 20 written practice tasks, worked solutions and 60 actuarial concepts.
Survival Models for Contingent Cash Flows
1. Mortality curve shape and age-specific hazards
A mortality curve describes how death risk varies with attained age, rather than imposing one constant risk throughout life. Typical features can include elevated infant mortality, a young-adult accident hump and increasing adult mortality. The force of mortality is an instantaneous conditional rate; an annual death probability integrates that rate over a year. Calendar period, birth cohort and selection can alter observed patterns. Use curve features to assess model suitability, while avoiding an assumption that every population exhibits the same shape or late-age behavior.
Worked example: Suppose mortality intensity is constant at 0.02 during one age-year. The annual death probability is 1-exp(-0.02), approximately 0.019801. Using 0.02 directly is a close approximation here, but it is not the exact probability.
Common mistake: Treating a force of mortality as an annual probability ignores the survival conditioning and integration that connect the two quantities.
Official syllabus reference: [2]
2. Frailty selection in heterogeneous populations
Population mortality can change because the composition of survivors changes. In a frailty mixture, higher-risk individuals tend to die sooner, increasing the proportion of lower-risk survivors. Aggregate mortality is therefore a survivor-weighted average of individual hazards, not generally an average using the original population weights. This selection mechanism can flatten the aggregate mortality curve even when each subgroup's hazard remains constant or increases. Apply mixture models when evaluating heterogeneity, and distinguish this composition effect from genuine improvement in an individual's underlying mortality.
Worked example: Start with equal numbers having constant annual hazards 0.01 and 0.05. After ten years, their surviving weights are proportional to exp(-0.10) and exp(-0.50). The aggregate hazard is approximately 0.0261, below the initial average of 0.03.
Common mistake: Keeping the initial subgroup weights throughout a projection overlooks selective survival and can overstate the later aggregate mortality rate.
Official syllabus reference: [2]
3. Markov state sufficiency and disability duration
A Markov model requires the current state and relevant time variables to contain the information needed for future transitions. Intensities may vary with attained age or calendar time without violating this property. However, recovery depending on time since disability onset is not captured by a single undifferentiated disabled state. A semi-Markov model explicitly retains duration since entry; an expanded Markov model can approximate it with duration substates. The practical decision is whether the chosen states preserve enough history to support the intended benefits and transition assumptions.
Worked example: Two disabled members have the same age but have been disabled for one month and three years. If their recovery prospects differ, assigning both the same age-dependent recovery intensity loses information. Separate short-duration and long-duration disability states can improve the approximation.
Common mistake: Believing that age-varying intensities violate the Markov property confuses permitted time dependence with dependence on omitted individual history.
Official syllabus reference: [2]
4. Kolmogorov forward equations as probability balances
Forward equations describe how destination-state probabilities change through incoming and outgoing flows. For a row probability vector p(t) and generator Q(t), the equation is p'(t)=p(t)Q(t). Off-diagonal generator entries are nonnegative transition intensities; each diagonal is minus the sum of its row's exit intensities. Consequently, generator rows sum to zero and total probability remains one. Derive equations from the transition diagram before calculating. Reversible transitions require inflows as well as outflows, so occupancy cannot generally be obtained from a single survival exponential.
Worked example: If healthy-to-disabled intensity is 0.04 and disabled-to-healthy intensity is 0.06, healthy occupancy receives the term 0.06p_D. With healthy death intensity 0.01, its derivative is -0.05p_H+0.06p_D, explicitly allowing recovery to replenish the healthy state.
Common mistake: Writing only exit terms treats every state as absorbing after departure and misses probability returning through recovery transitions.
Official syllabus reference: [2]
5. Transition probabilities from a generator
For constant generator Q, the transition matrix over duration t is exp(tQ), the matrix exponential. It includes paths involving multiple transitions, such as disability followed by recovery. The approximation I+hQ is a first-order Euler step, useful for short intervals but not exact. Large steps can yield negative diagonal probabilities. For piecewise constant intensities, calculate each interval's transition matrix and multiply them in chronological order. Validate nonnegativity, row sums and convergence when using numerical approximations, especially when transitions are rapid relative to the chosen time step.
Worked example: A state has total exit intensity 0.80 per year. An Euler step of two years gives a diagonal entry 1-1.60=-0.60, which cannot be a probability. Shorter steps or matrix exponentiation avoid this particular approximation failure.
Common mistake: Exponentiating each generator entry separately does not produce a transition matrix; the matrix exponential accounts for interacting transition pathways.
Official syllabus reference: [2]
6. Chapman-Kolmogorov composition across intervals
The Chapman-Kolmogorov equation combines transition probabilities over consecutive intervals by summing over the intermediate state. In matrix form, P(s,u)=P(s,t)P(t,u). The first matrix describes the earlier interval, so order matters when matrices differ. This method supports discrete-time Markov models and continuous-time models evaluated at selected times. It does not assume transition matrices are identical across ages. Its validity depends on the state at the intermediate time being sufficient for subsequent transition probabilities, consistent with the Markov representation being used.
Worked example: Suppose one-year healthy retention is 0.90 and healthy-to-disabled probability is 0.08. In the next year, healthy retention is 0.85 and disabled-to-healthy recovery is 0.20. Two-year healthy probability includes 0.90×0.85+0.08×0.20=0.781, before any other possible intermediate states.
Common mistake: Multiplying only same-state probabilities excludes routes through other intermediate states and can understate occupancy when recovery is possible.
Official syllabus reference: [2]
7. Competing hazards and associated single decrements
In a multiple-decrement model, each cause acts on lives still exposed to all competing causes. With constant intensities μ_j, total survival is exp(-tΣμ_j), and cause-j decrement probability is μ_j[1-exp(-tΣμ_j)]/Σμ_j. An associated single-decrement model removes the other causes while retaining the specified cause's intensity, giving 1-exp(-tμ_j). Observed and associated probabilities therefore differ. Constructing a table requires consistent assumptions about how causes operate within each age interval, rather than simply adding probabilities taken from independently constructed single-decrement tables.
Worked example: With death intensity 0.02 and withdrawal intensity 0.03, one-year observed death probability is 0.4[1-exp(-0.05)]=0.019508. The associated death-only probability is 1-exp(-0.02)=0.019801 because withdrawals no longer remove lives before a possible death.
Common mistake: Adding associated single-decrement probabilities treats competing exits as if they could each act on the full original population throughout the year.
Official syllabus reference: [2]
8. UDD assumptions when reconstructing decrement tables
Fractional-age assumptions determine how annual decrement probabilities translate into within-year survival. Under uniform distribution of decrements in an associated single-decrement model, survival to fraction t is 1-tq'. If two associated decrement times are modeled independently, observed cause-one probability is the integral of q1'(1-tq2') over the year, giving q1'(1-q2'/2). This construction differs from assuming decrements are uniform directly in the multiple-decrement table. State where UDD is imposed before converting probabilities; changing that location changes the resulting relationships.
Worked example: Associated annual probabilities are 0.10 and 0.20, with independent associated decrement times and UDD in each model. Observed cause-one probability is 0.10×0.90=0.09, cause-two probability is 0.20×0.95=0.19, and total survival is 0.90×0.80=0.72.
Common mistake: Using a UDD conversion formula without identifying its underlying single-decrement or multiple-decrement assumption can produce an internally inconsistent table.
Official syllabus reference: [2]
9. Transition-intensity maximum likelihood from exposure
For completely observed Markov paths with constant intensities within an interval, the relevant log-likelihood contains d_ij log μ_ij-E_iμ_ij for each transition type. Here d_ij is the observed transition count and E_i is person-time exposed in the origin state. Maximization gives μ̂_ij=d_ij/E_i. Repeated transitions by one person can contribute multiple events and exposure segments. Right censoring contributes exposure up to censoring, provided the observation mechanism is appropriately noninformative. Grouped observations without full paths require a different likelihood based on transition probabilities.
Worked example: An insurer records 24 recoveries during 160 disabled person-years. The estimated recovery intensity is 24/160=0.15 per year. Dividing by healthy exposure would estimate no meaningful recovery rate because healthy members are not exposed to the recovery transition.
Common mistake: Using the number initially enrolled as the denominator ignores unequal observation time and time spent outside the transition's origin state.
Official syllabus reference: [2]
10. Asymptotic confidence intervals for transition estimates
For a constant-intensity estimate d/E, observed information gives approximate standard error sqrt(d)/E. A Wald interval is μ̂±z×SE, relying on a sufficiently informative sample and asymptotic normality. It can extend below zero when counts are small, signaling poor approximation. For an independent binomial transition-probability estimate p̂=d/n, the familiar standard error is sqrt[p̂(1-p̂)/n], under that sampling model. Do not interchange these formulas: event counts over person-time and one-transition categorical observations have different likelihoods and denominators.
Worked example: There are 36 disability transitions over 900 active person-years. Estimated intensity is 0.04 and standard error is 6/900=0.006667. A 95% Wald interval is approximately 0.02693 to 0.05307 per year, using the stated normal approximation.
Common mistake: A narrow interval reflects estimated sampling precision under the model; it does not establish that the model or observation process is unbiased.
Official syllabus reference: [2]
Premium and Policy Valuation for Long-Term State-Dependent Coverages
11. State-contingent present-value random variables
A present-value random variable records discounted payments along an individual policy path. For scheduled benefits, write Z=Σv(t_k)b_k I{X(t_k) is eligible}. Its expectation replaces indicators with occupancy probabilities, without requiring independence between payment dates. Variance is different: it generally requires joint probabilities because payments on the same path are dependent. Define whether payments occur before or after state transitions at a shared date. This path-based formulation helps distinguish payment conditions from the probability model used to evaluate them.
Worked example: A contract pays 500 at years one and two if disabled. With discount factor 0.95 per year and disabled probabilities 0.10 and 0.18, EPV is 500[0.95×0.10+0.95²×0.18]=128.725, regardless of dependence between the two indicators.
Common mistake: Using marginal occupancy probabilities to calculate a second moment ignores covariance and can substantially misstate variability of the present value.
Official syllabus reference: [2]
12. Continuous state annuities as discounted occupancy
A continuous state annuity pays a rate while a specified state is occupied. Its EPV is the integral of the payment rate times discounting times state occupancy probability. Occupancy may reflect repeated entries and recoveries, so it is not necessarily the probability of having entered the state at least once. The payment rate is measured per unit time, making integration necessary. This representation applies to disability income, care costs and CCRC maintenance expenses, with different rates assigned to different states and durations.
Worked example: Assume disabled occupancy is a constant 0.20 over the next two years, payment rate is 6,000 annually and interest force is 0.04. EPV is 1,200[1-exp(-0.08)]/0.04≈2,306.51, representing discounted expected time spent receiving benefits.
Common mistake: Multiplying the benefit rate by disability incidence counts entries instead of benefit-paying time and misses the duration of each disability spell.
Official syllabus reference: [2]
13. Transition benefits and occupancy-intensity integrals
A benefit paid immediately on transition i→j is valued using the probability of occupying origin state i multiplied by transition intensity μ_ij and the benefit amount. Integrating this product with discounting gives the expected present value over the coverage term. The origin-state probability must include all permitted routes into that state. If transitions can recur, the integral counts each qualifying event unless the contract or state structure limits payment. A benefit paid on entry is therefore different from an annuity payable while remaining in the destination state.
Worked example: A five-year contract pays 2,000 on each healthy-to-disabled transition. Its EPV is the integral from zero to five of v(t)×2,000×p_H(t)×μ_HD(t) dt. Recovery and subsequent redisability can generate additional payments if the contract explicitly permits them.
Common mistake: Weighting an entry benefit by destination occupancy can pay for long-standing cases and fails to isolate the transition that triggers the cash flow.
Official syllabus reference: [2]
14. Woolhouse corrections and payment normalization
Woolhouse formulas approximate frequent-payment annuity factors from annual factors by correcting payment timing. For a whole-life annuity-due normalized to annual total one, the two-term correction subtracts (m-1)/(2m). A common three-term form also subtracts (m²-1)/(12m²). This local correction uses the initial decline in discounted survival. Temporary and more general state-dependent payments can require endpoint or derivative terms, so the whole-life expression should not be transferred indiscriminately. Always identify whether the factor values payments of one or payments of 1/m.
Worked example: If the annual whole-life annuity-due factor is 12 and payments become quarterly at one quarter each, the two-term approximation is 12-3/8=11.625. An annual pension rate of 8,000 then has approximate EPV 93,000, not 23,250.
Common mistake: Applying the whole-life third-term formula unchanged to a temporary annuity omits terminal effects that can matter for the approximation.
Official syllabus reference: [2]
15. Equivalence premiums with proportional expenses
The equivalence principle sets expected premium income equal to expected benefits and expenses on a specified valuation basis. When expenses include a percentage of premiums, that expense is itself a function of the unknown premium. Bring proportional expenses to the premium side before solving. Premium eligibility may differ from benefit eligibility, especially under disability waiver or state-dependent charges. The result is a zero expected initial loss premium on the chosen assumptions, rather than a profit-loaded price or a guarantee against unfavorable individual experience.
Worked example: Benefit EPV is 18,000, fixed expense EPV is 600 and the premium annuity factor is 8. If expenses equal 5% of every premium, 0.95×8P=18,600, giving annual premium approximately 2,447.37 under the stated payment schedule.
Common mistake: Adding a percentage expense after solving the net premium generally understates the gross premium because the expense also applies to the added premium.
Official syllabus reference: [2]
16. CCRC entry fees and state-dependent service costs
A CCRC valuation can assign different ongoing costs to independent living, assisted living and nursing care, with fees depending on the contract. Expected cost is the sum of discounted occupancy-weighted service rates plus any transition or refund benefits. An entry fee is a time-zero cash flow, while maintenance fees form a state-contingent annuity. Refunds may depend on death, exit or elapsed residence. Solve the equivalence equation using the actual fee and refund design; an average annual cost alone can conceal important timing and care-state differences.
Worked example: Expected service-cost EPV is 240,000 and refund EPV is 20,000. Maintenance fees of 12,000 have an eligibility annuity factor of 10. With no other cash flows, the equivalence entry fee is 240,000+20,000-120,000=140,000.
Common mistake: Treating a refundable entry fee entirely as unrestricted income overlooks the future refund liability and can materially underprice the contract.
Official syllabus reference: [2]
17. Prospective reserves conditional on current state
A prospective reserve is the conditional expected present value of future benefits and expenses less future premiums, given the policy's current information and state. Different current states generally require different reserves because future payment and transition patterns differ. Retrospective accumulation describes past cash flows instead; equality with a prospective expression requires compatible assumptions and conventions. Specify valuation timing relative to premiums and benefits. The reserve is an expected liability on its basis, not the realized future loss of every policy or necessarily its surrender value.
Worked example: At a valuation date, future benefit and expense EPV is 30,000 for a disabled policy and future premium EPV is zero under waiver. Its prospective reserve is 30,000. A healthy policy with identical age may have a different reserve because premiums continue and transitions differ.
Common mistake: Assigning every policy of the same age the same reserve ignores current-state information that directly changes expected future cash flows.
Official syllabus reference: [2]
18. Thiele's equation and transition reserve strain
With premium rate p_i, benefit outgo rate b_i and transition payment B_ij, Thiele's equation is V_i'=δV_i+p_i-b_i-Σμ_ij(B_ij+V_j-V_i). The bracket is the transition strain: immediate payment plus destination reserve less released origin reserve. A transition with no cash payment can still create a gain or loss through reserve transfer. The equation follows from balancing expected resources over a short interval. Use consistent signs and units, and solve with appropriate boundary conditions rather than interpreting the differential equation as a complete valuation by itself.
Worked example: Suppose disability entry pays 3,000, healthy reserve is 10,000 and disabled reserve is 40,000. Entry strain is 33,000. At intensity 0.02, this transition contributes -660 per year to the healthy reserve's forward derivative, holding other terms fixed.
Common mistake: Using only the immediate transition benefit omits the destination liability and can severely understate the financial strain of entering care.
Official syllabus reference: [2]
19. Reserve jumps and terminal boundary conditions
A reserve immediately before a scheduled payment differs from the reserve immediately afterward. If benefit B is paid and premium P is received at the same instant, a consistent convention gives V_before=B-P+V_after, conditional on the relevant state and no intervening uncertainty. At contract termination without remaining obligations, the post-payment reserve is zero. A maturity benefit therefore creates a nonzero pre-payment terminal reserve. These jump conditions connect continuous Thiele solutions across discrete cash flows and prevent premiums or benefits from being omitted or counted twice.
Worked example: A surviving policy pays a maturity benefit of 20,000 at time ten and then ends. Its reserve immediately after payment is zero, but immediately before payment it is 20,000. Setting the pre-payment terminal reserve to zero would omit the maturity obligation.
Common mistake: Applying a zero terminal reserve without specifying whether final payments have occurred can erase a contractual benefit from the valuation.
Official syllabus reference: [2]
Joint Life Insurance and Annuities
20. First-death and last-death status lifetimes
For future lifetimes T_x and T_y, joint-life status fails at min(T_x,T_y), while last-survivor status fails at max(T_x,T_y). These random variables determine whether insurance pays at first or second death and whether an annuity continues while both or either remain alive. Their distributions can be derived from joint survival probabilities without assuming independence. A status model can summarize the timing of failure, whereas a multiple-state model also identifies which survivor remains and can support survivor-specific payments or changing mortality after bereavement.
Worked example: If one spouse dies after four years and the other after nine, joint-life status lasts four years and last-survivor status lasts nine. A first-death insurance pays at the earlier event, while a second-death insurance waits for the later event.
Common mistake: Interchanging minimum and maximum lifetimes changes both the payment trigger and duration of cover, even when the same two lives are insured.
Official syllabus reference: [2]
21. Joint-survival identities and dependence boundaries
Let S_x(t), S_y(t) and S_xy(t) denote marginal and joint survival. Last-survivor survival is S_x+S_y-S_xy by inclusion-exclusion, regardless of independence. Independence additionally permits S_xy=S_xS_y. With marginals fixed, dependence changes the joint term and therefore changes both joint-life and last-survivor values. The joint probability must lie between max(0,S_x+S_y-1) and min(S_x,S_y). These bounds provide useful checks when a proposed dependence model or supplied probability appears inconsistent with the marginal survival assumptions.
Worked example: If marginal survival probabilities are 0.80 and 0.70, joint survival must lie between 0.50 and 0.70. Independence gives 0.56, producing last-survivor survival 0.94. A proposed joint survival of 0.75 is incompatible with the second marginal.
Common mistake: Using the product of marginal survival probabilities without an independence assumption silently fixes the dependence structure and can distort benefit values.
Official syllabus reference: [2]
22. Common-shock mortality in joint-life models
A common-shock model includes a transition from both alive directly to both dead, alongside individual-death transitions. With constant individual intensities λ_x and λ_y and common-shock intensity λ_c, both-alive survival is exp[-(λ_x+λ_y+λ_c)t]. In the standard independent-shock construction, marginal mortality intensities include λ_c as well. Simultaneous death creates dependence and affects first-death and second-death benefits differently. When valuing survivor benefits, common shocks reduce exposure in both-alive status but do not create a one-survivor state.
Worked example: Individual shock intensities are 0.01 and 0.02, and common-shock intensity is 0.005. Both-alive exit intensity is 0.035, while marginal death intensities are 0.015 and 0.025. Adding the marginals gives 0.040 and double-counts the common shock for first exit.
Common mistake: Adding marginal death intensities to obtain first-death intensity double-counts simultaneous-death risk when both marginals already include a common shock.
Official syllabus reference: [2]
23. Bereavement effects through survivor-state mortality
Dependence can arise because one death changes the surviving partner's mortality. A multiple-state model represents this with separate both-alive and survivor states, assigning different mortality intensities after bereavement. If the effect depends on time since bereavement, a single survivor state may be insufficient; duration tracking or additional substates may be needed. This mechanism differs from a common shock, which kills both simultaneously. It is especially relevant for reversionary annuities because the payment period begins precisely when the survivor enters the potentially altered mortality state.
Worked example: A spouse has mortality intensity 0.02 while both partners live, rising to 0.05 after the partner dies. Valuing a survivor pension using 0.02 throughout overstates expected survivor payment duration under these assumptions, even if both-alive probabilities were calculated correctly.
Common mistake: Treating all dependence as simultaneous death overlooks changes in the surviving partner's later mortality and their effect on reversionary payments.
Official syllabus reference: [2]
24. Joint-life premiums with status-dependent payment terms
A joint-life premium follows the equivalence principle, but its denominator must reflect the status under which premiums are payable. Premiums may stop at first death even when the benefit is payable at second death. Such a contract combines a last-death benefit EPV with a joint-life premium annuity. Alternatively, premiums may continue while either life survives. Identify both triggers separately before dividing benefit and expense EPV by the premium annuity factor. Dependence assumptions influence the numerator and denominator through potentially different combinations of state probabilities.
Worked example: A second-death benefit has EPV 40,000. Annual premiums stop at the first death, and the applicable joint-life premium annuity factor is 8. With no expenses, premium is 5,000. A last-survivor factor of 11 would apply only to a different premium condition.
Common mistake: Matching the premium annuity automatically to the benefit's death order ignores contracts whose premiums and benefits depend on different life statuses.
Official syllabus reference: [2]
25. Reversionary annuities from survivor-state occupancy
A reversionary annuity to y after x dies pays when y is alive and x is dead. At each payment time, that event has probability S_y(t)-S_xy(t). Therefore, with matching payment timing and amounts, its annuity factor equals the single-life annuity on y minus the joint-life annuity. This identity does not require independence. A state model may be needed to calculate the underlying probabilities under bereavement dependence. Distinguish this benefit from an annuity payable while either life survives, which includes the both-alive state.
Worked example: At a scheduled pension date, y survives with probability 0.85 and both survive with probability 0.72. The reversionary payment probability is 0.13. A payment of 2,000 discounted by 0.90 contributes 234 to the benefit EPV at that date.
Common mistake: Using last-survivor probability for a reversionary pension incorrectly includes payments while both lives remain alive and the pension has not yet commenced.
Official syllabus reference: [2]
26. Joint-life reserves after the first death
A joint-life contract's reserve must be conditioned on its current survival state. Before first death, future cash flows may include several branches; afterward, only the surviving-life obligations remain. Premium cessation, immediate first-death payments and survivor pensions can all create reserve jumps or state transfers. Apply the same prospective definition separately in both-alive, x-only and y-only states. A reserve averaged across these states may be useful for a portfolio projection, but it is not the conditional reserve appropriate to a known surviving couple's state.
Worked example: A contract pays a survivor pension only after the first death and collects premiums only while both live. Immediately after x dies, future premiums cease and y's pension begins. The y-only reserve must value that pension rather than reuse the earlier both-alive reserve.
Common mistake: Continuing to subtract both-alive future premiums after one insured dies understates the survivor-state liability when those premiums cease at first death.
Official syllabus reference: [2]
Profit Analysis
27. Expected versus actual annual surplus
Annual surplus compares available assets with claims, expenses and closing liabilities on a consistent policy-count basis. Expected surplus uses assumed mortality, expenses and investment returns; actual surplus substitutes realized experience while retaining the stated reserve convention. Opening reserves are resources released into the year, and closing reserves absorb resources for surviving obligations. Timing matters because premiums and expenses may earn or forgo interest. A favorable cash balance is not automatically profit if substantial closing liabilities remain, and portfolio totals must not be compared directly with per-survivor amounts.
Worked example: Opening reserve 200 plus premium 100 less immediate expense 10 leaves 290 invested. At 5%, assets become 304.50. Expected claims of 20 and expected closing reserve cost of 260 produce expected surplus 24.50 per opening policy.
Common mistake: Ignoring the closing reserve treats funds needed for future benefits as distributable profit and overstates the year's economic emergence.
Official syllabus reference: [2]
28. Mortality gain and net death strain
When death terminates a policy and releases its reserve, mortality gain depends on the net death strain rather than the gross death benefit. In a simple year-end model, strain is death benefit minus the reserve that would otherwise be held for a survivor. Lower mortality can create a loss for annuities or contracts with negative death strain. Gain-by-source decompositions should specify the order in which experience assumptions replace expected assumptions, because interaction terms may otherwise be allocated inconsistently. The sign must follow the actual cash flows and reserve release.
Worked example: Death benefit is 1,000 and survivor reserve is 700, giving strain 300. If actual deaths exceed expected deaths by ten, mortality loss is approximately 3,000 under fixed timing and other assumptions, rather than the 10,000 increase in gross claims.
Common mistake: Assuming lower mortality always improves profit overlooks annuity longevity risk and contracts where reserve release exceeds the death payment.
Official syllabus reference: [2]
29. Conditional profit vectors and issued-policy signatures
Profit testing distinguishes profit conditional on a policy being in force from expected profit per policy originally issued. To convert a beginning-year conditional profit into an issued-policy cash flow, multiply by the probability of being in force at that year's beginning. Terminology for vector and signature can vary, so define the basis explicitly. Do not apply survival again if the conditional year's profit already incorporates that year's death and lapse outcomes. This distinction is essential before discounting or comparing products with different persistency patterns.
Worked example: Conditional profit in year four is 80 per policy present at its beginning. If 70% of issued policies reach that point, expected year-four profit per issued policy is 56. Multiplying by another year-end survival factor would double-adjust if already included within the 80.
Common mistake: Discounting conditional profits without in-force weighting overstates later issued-policy earnings when death or lapse has reduced the portfolio.
Official syllabus reference: [2]
30. NPV and premium-based profit margin
Profit NPV discounts the expected issued-policy profit cash flows at a specified risk discount rate, including any issue-time strain. A premium-based profit margin divides that NPV by the present value of expected premiums using the stated convention. Numerator and denominator require consistent policy-count and timing bases. Margin is not the same as profit divided by undiscounted premium or investment return on capital. A positive margin is meaningful only relative to the chosen assumptions, discount rate and included costs, particularly acquisition expense and guarantee funding.
Worked example: Issue strain is 100, followed by expected profits of 70 after one and two years. At 10%, NPV is -100+70/1.10+70/1.10²=21.49. If premium PV is 500, the corresponding premium-based profit margin is approximately 4.30%.
Common mistake: Using undiscounted premiums beneath a discounted profit numerator changes the margin definition and makes comparisons inconsistent with the specified objective.
Official syllabus reference: [2]
31. Internal rate of return and cash-flow sign changes
The internal rate of return is a discount rate at which profit NPV equals zero. For a conventional pattern with an initial outflow followed by positive inflows, the economically relevant nonnegative root is usually straightforward to identify. Multiple sign changes can produce multiple roots or make IRR an unreliable ranking measure. Numerical solving should therefore include a cash-flow inspection and a root bracket. NPV at the required hurdle rate directly measures value on that basis and remains useful when IRR interpretation becomes ambiguous.
Worked example: An initial outflow of 100 followed by 60 at each of the next two year-ends has NPV zero when 100(1+r)²=60(1+r)+60. The positive solution is approximately 13.07%, which can be checked by substitution into the discounted cash flows.
Common mistake: Accepting the first spreadsheet IRR result without checking sign changes can conceal additional roots or dependence on the solver's initial guess.
Official syllabus reference: [2]
32. Discounted payback and crossing-year conventions
Discounted payback identifies when cumulative discounted profit first recovers the initial strain. Calculate discounted issued-policy profits, accumulate them chronologically and find the first nonnegative cumulative total. If all modeled profits occur at year-end, the literal payback date is that year-end. Fractional-year interpolation adds an assumption about profit emergence within the crossing year. Payback ignores profits after recovery and therefore does not measure total value. It is most informative alongside NPV and margin, especially when comparing products with different acquisition strain and late-duration earnings.
Worked example: Cumulative discounted profit is -12 after year two, and year-three discounted profit is 30. Year-end payback is year three. Under an explicit uniform-emergence interpolation, fractional payback is 2+12/30=2.4 years; the interpolation is an added convention.
Common mistake: Reporting an interpolated payback as an exact cash-flow date hides the assumption that profit emerges continuously within an otherwise annual model.
Official syllabus reference: [2]
33. Premium solving under an explicit profit objective
Profit-based pricing chooses a premium that meets a specified objective, such as NPV, margin or IRR, rather than merely equating benefit and premium EPVs. Build all premium-dependent cash flows first, including percentage expenses, commissions, account credits and reserves. If NPV is linear in premium, solve algebraically; otherwise use a numerical root method with feasibility checks. A margin target places premium in both numerator and denominator. The resulting price is conditional on the complete profit-test basis, so changing reserves or discount rates can change the answer.
Worked example: Suppose profit NPV is 6P-900 and premium PV is 8P. A target margin of 10% requires 6P-900=0.10×8P. Thus 5.2P=900 and P≈173.08, rather than the zero-NPV premium of 150.
Common mistake: Adding the target margin as a simple percentage markup on a net premium ignores its defined denominator and premium-related expenses.
Official syllabus reference: [2]
34. Reserve zeroization by backward profit testing
Reserve zeroization uses projected future profits to establish reserves that prevent specified negative profit emergence, subject to the chosen method and constraints. Work backward from a terminal condition, adding reserve where future resources would otherwise be insufficient. Merely lowering a current reserve can move a loss to a later year rather than eliminate it. Reserve release, reserve interest and policy termination must all be reflected consistently. The exercise is a cash-flow funding procedure on a stated basis; it does not independently establish compliance with an unspecified regulatory valuation regime.
Worked example: A final-year policy needs 120 at year-end but receives net premium 100 at its beginning. At 5%, opening reserve R satisfies (R+100)1.05=120, so R=14.2857. That required reserve then becomes part of the preceding year's closing funding requirement.
Common mistake: Removing a negative profit by reducing the next reserve without projecting later years can simply defer the shortfall beyond the period being inspected.
Official syllabus reference: [2]
35. Profit-test sensitivities and interacting stresses
A sensitivity changes one assumption to identify its effect on profit measures, while a scenario can change several assumptions together. Mortality, lapse, expenses and investment returns interact through policy counts, reserves and cash-flow timing. Therefore combined stress effects need not equal the sum of isolated effects. Preserve contractual guarantees while stressing experience assumptions; a guaranteed credited rate cannot be reduced merely because asset returns fall. Interpret both total NPV and the timing of losses, since products with similar stressed NPV can require very different near-term funding.
Worked example: A lapse stress may reduce future fee income while also accelerating surrender payments. Combining it with low returns can create a larger loss than either isolated stress because fewer remaining accounts generate future margins to recover the early shortfall.
Common mistake: Adding independent sensitivity losses mechanically assumes away interactions that may be central to the product's guarantees and persistency-dependent recovery of acquisition costs.
Official syllabus reference: [2]
Pension Plans and Retirement Benefits
36. FAS, CAE and CARE benefit construction
A final-average-salary benefit applies an accrual rate and service to salary measured near retirement. A career-average-earnings benefit uses earnings averaged across the career under the plan's definition. A CARE benefit accumulates annual pension credits, revaluing earlier credits according to a specified rule. The revaluation rate may differ from salary growth and post-retirement pension increases. Calculate each year's credit before applying its relevant revaluation period. Replacement ratios also require an explicit salary denominator, which need not be the salary measure used inside the benefit formula.
Worked example: A CARE plan credits 2% of salary each year. A 40,000 salary creates pension credit 800. If that credit receives two years of 3% revaluation before retirement, it contributes 800×1.03²=848.72 to annual pension, before adding other service years.
Common mistake: Revaluing the entire CARE total for the same number of years overstates newer credits, which have had less time to accumulate revaluation.
Official syllabus reference: [2]
37. DC fund recursion and contribution timing
A DC account accumulates contributions and investment returns according to their timing. With end-year contribution C_t and annual return i_t, the recurrence is F_t=F_(t-1)(1+i_t)+C_t, before any separately specified charges. Beginning-year contributions instead earn that year's return. Salary-related contributions require a salary projection consistent with the contribution date. Retirement income then depends on how the final fund is converted into benefits, so the accumulation return and retirement annuity price are separate assumptions. Deterministic calculations illustrate funding mechanics without eliminating market or longevity uncertainty.
Worked example: An opening fund of 10,000 earns 5%, and a contribution of 2,000 is paid at year-end. Closing fund is 12,500. If the contribution were paid at the beginning, closing fund would be 12,600 because the contribution also earns 5%.
Common mistake: Using a beginning-year accumulation formula for end-year contributions gives every contribution an extra year of return and understates required saving.
Official syllabus reference: [2]
38. DC contribution rates for target replacement income
To solve a DC contribution target, first multiply target replacement ratio by the specified retirement salary measure. Convert that income into required capital using the retirement annuity price. Subtract the accumulated value of existing assets, then divide the remaining capital need by the accumulated value of one unit of contribution rate applied to projected salaries. A negative result indicates existing assets exceed the deterministic target, subject to any contribution constraints. The calculation is sensitive to returns, salary growth, retirement timing and annuity pricing, so its assumptions must remain explicit.
Worked example: Target annual pension is 20,000 and annuity price is 15, requiring capital 300,000. Existing assets project to 240,000. If the accumulated salary base for future contributions is 400,000, the required constant contribution rate is 60,000/400,000=15%.
Common mistake: Dividing the capital shortfall by undiscounted salary totals ignores the investment accumulation earned by contributions before retirement and misstates the contribution rate.
Official syllabus reference: [2]
39. Salary scales as relative earnings functions
A salary scale expresses relative salary levels by age or service, often separated from general earnings inflation. If current salary at age x is S_x and the applicable scale is s, projected salary at y can be S_x(s_y/s_x), with additional escalation applied only if not already included. The ratio matters; the arbitrary normalization of the scale does not. Use the projected salaries required by the benefit formula, such as a final averaging window rather than one terminal salary when those differ. Avoid double-counting inflation embedded in a supplied scale.
Worked example: Current salary is 50,000, with scale values 1.20 now and 1.50 at retirement. A scale already including all assumed growth projects retirement salary as 50,000×1.50/1.20=62,500. Applying an additional inflation factor would duplicate growth under that specification.
Common mistake: Multiplying salary by the future scale level alone ignores its current normalization and can produce different answers for equivalent rescaled functions.
Official syllabus reference: [2]
40. Service-table probabilities for pension pathways
A service table tracks active membership under competing exits such as retirement, withdrawal, disability and death. Ratios of active counts can represent probabilities of remaining active when the table is constructed on that basis. Exit counts or probabilities identify the pathway triggering a particular pension benefit. Remaining active to retirement differs from merely surviving to retirement, especially when deferred benefits vest on withdrawal. Match each benefit to the correct pathway and apply any subsequent survival probability separately. Multiple-decrement assumptions govern how the competing exits share exposure.
Worked example: A service table has 1,000 active members at age 50 and 700 at age 60. The implied probability of remaining active is 0.70. A vested deferred pension payable after withdrawal cannot automatically use that factor because employment continuation is no longer required.
Common mistake: Replacing active-service survival with ordinary life survival ignores withdrawal and other employment exits that can change benefit eligibility and payment amounts.
Official syllabus reference: [2]
41. Vested deferred benefits on early exit
Early exit can freeze an earned DB pension, create a deferred pension subject to revaluation, or produce another specified benefit. Start with the service and salary recognized at exit, then apply vesting, revaluation and retirement-age provisions explicitly. Valuation after exit generally requires survival to commencement rather than continued active membership. Early-retirement reductions and deferred-benefit revaluation are different mechanisms: one adjusts commencement terms, while the other changes the benefit before commencement. No universal vesting or reduction schedule should be assumed when the contract has not supplied it.
Worked example: A departing member has eight years of recognized service, salary 50,000 and accrual rate 1.5%. The frozen annual pension is 6,000. If the plan explicitly revalues it by 2% for ten years, commencement pension becomes approximately 7,313.97.
Common mistake: Projecting post-exit salary as though employment continued can overstate a frozen deferred benefit when the plan recognizes only salary at departure.
Official syllabus reference: [2]
42. Projected unit credit and salary projection
Projected unit credit allocates projected retirement benefit to service units while incorporating the salary projection required by the plan. For a simple final-salary formula, accrued liability values completed-service units using projected final salary; normal cost values the unit attributed to the coming service year under the stated timing convention. Survival, employment eligibility, discounting and annuity conversion are then applied. PUC does not assign all future service to the accrued liability. Its distinction is projecting the benefit attached to earned service, rather than treating future service itself as already accrued.
Worked example: Completed service is ten years, accrual rate 2%, projected final salary 80,000 and retirement-value factor 5. The PUC accrued liability is 10×0.02×80,000×5=80,000. Under a one-unit convention, normal cost is 8,000 for the additional service year.
Common mistake: Including every expected future service year in the PUC accrued liability confuses total projected benefit value with the portion allocated to past service.
Official syllabus reference: [2]
43. Traditional unit credit and salary-related catch-up
Traditional unit credit values the benefit accrued using salary recognized at the valuation date, rather than projecting future salary for existing service units. New service adds another unit, while later salary increases can raise the value of earlier units in a final-salary plan. Those salary-related increases must be reflected in a complete liability roll-forward; they do not disappear because initial normal cost used current salary. Exact normal-cost timing and treatment of expected salary changes should follow the stated valuation convention. Compare TUC and PUC on otherwise consistent assumptions.
Worked example: A member has ten service years and a 2% accrual rate. Salary rises from 50,000 to 55,000. The annual pension attributed to those existing years rises from 10,000 to 11,000, creating an additional 1,000 of annual accrued benefit before new service is counted.
Common mistake: Assuming TUC permanently freezes earlier salary units overlooks the final-salary linkage and understates future liability growth when salary increases occur.
Official syllabus reference: [2]
44. Pension liability roll-forwards and experience gains
A liability roll-forward explains changes through interest, service accrual, benefit payments, membership movements and assumption or experience differences. The exact ordering depends on whether normal cost and contributions are measured at the beginning or end of the year. Contributions affect plan assets and funding, but do not directly erase the actuarial liability for promised benefits. Compare actual closing liability with the expected closing liability on a consistent method to identify gains or losses. Salary experience, retirement timing and survival can affect liabilities differently under PUC and TUC.
Worked example: Opening liability is 100 and beginning-year normal cost is 10. With 5% interest and year-end benefits of 8, expected closing liability is (100+10)1.05-8=107.50. A separately calculated actual closing liability of 112 implies an adverse difference of 4.50.
Common mistake: Subtracting employer contributions from the benefit liability confuses funding assets with the present value of obligations owed to members.
Official syllabus reference: [2]
Universal Life Insurance
45. Type A benefits and declining net amount at risk
Under a basic Type A UL design, total death benefit is fixed while account value finances part of that amount. The insurer's net amount at risk is therefore death benefit less the account value used by the contract's COI convention, subject to any stated floors or corridors. As the account grows, net risk generally falls while the total death payment stays level. Account deductions and credited interest still affect sustainability. Do not assume additional corridor rules or tax constraints unless supplied; these can alter the simple fixed-benefit relationship.
Worked example: A Type A policy has fixed death benefit 100,000 and pre-COI account 25,000. Net amount at risk is 75,000. At a stipulated annual COI rate of 0.008, the charge is 600 before other deductions or interest are applied.
Common mistake: Adding account value to the stated fixed Type A death benefit converts the calculation into a different benefit design and overstates the death payment.
Official syllabus reference: [2]
46. Type B benefits and level additional insurance
A basic Type B UL design pays a specified face amount plus account value at death. The account component is funded by the account itself, leaving the face amount as the insurer's net amount at risk under the matching timing convention. Account growth therefore increases the total death benefit instead of reducing the additional insurance. Compared with Type A, this can produce higher COI for the same face amount and account. Distinguish the specified face amount from total benefit, and use the contract's actual measurement dates for the account component.
Worked example: A Type B policy has face amount 100,000 and account 25,000, giving death benefit 125,000. Net risk is 100,000. At COI rate 0.008, the charge is 800, compared with 600 for a matching fixed-benefit Type A example.
Common mistake: Subtracting account value from the Type B face amount understates net insurance risk because the account is paid in addition to that face amount.
Official syllabus reference: [2]
47. UL account roll-forwards and deduction order
A UL account roll-forward applies premium allocation, policy charges, COI and credited interest in the contractual sequence. A common stipulated form is A_t=A_(t-1)+(1-load)P-expense-COI, but other sequences are possible. Charges based on account value may change when timing changes, making order economically important. Keep credited interest distinct from the insurer's earned investment return. If deductions exhaust the account, the ordinary recursion may require an exhaustion rule or guarantee provision rather than allowing an unexplained negative policyholder account.
Worked example: Opening account is 5,000, premium 1,000, load 4%, expense 60 and COI 100. With all deductions before 3% interest, closing account is (5,000+960-60-100)1.03=5,974. The formula explicitly states which amounts receive interest.
Common mistake: Applying interest before beginning-period deductions credits interest on money already removed and overstates the account under the specified deduction-first convention.
Official syllabus reference: [2]
48. COI charges versus expected mortality cost
A cost-of-insurance charge is a contractual deduction, not necessarily the insurer's expected mortality cost. The charge rate may include margins and may be measured on a different timing basis from expected claims. Profit analysis compares COI income with the expected net death strain, including account or reserve release, rather than with gross death benefits alone. Mortality assumptions used for pricing experience can differ from charge rates guaranteed or currently applied. Keep these quantities separate to identify mortality margin without mistaking policyholder deductions for realized insurance claims.
Worked example: Net amount at risk is 80,000. A charge rate of 0.010 produces COI income 800, while assumed death probability 0.007 implies expected net mortality cost 560 under matching timing. The resulting mortality margin is 240 before expenses and other adjustments.
Common mistake: Treating the COI rate as the best-estimate death probability automatically erases any mortality margin embedded in the contractual charge schedule.
Official syllabus reference: [2]
49. No-lapse guarantees beyond account exhaustion
A no-lapse guarantee can preserve specified coverage after ordinary account mechanics would exhaust the policy, provided contractual qualifying conditions are met. Its liability depends on future guaranteed benefits, qualifying premiums and expenses, net of any account resources available under the design. A shadow account or premium test may determine eligibility without being an actual withdrawable account. Model the guarantee's conditions explicitly and distinguish account value from the reserve for additional insurer obligations. Zero account value can coexist with substantial remaining death cover and therefore a positive guarantee reserve.
Worked example: An exhausted account still supports one guaranteed coverage year. Expected death claims are 100, expenses 20 and qualifying premium 90, all valued on a zero-interest basis. The additional prospective funding requirement is 30 even though the policyholder's account is zero.
Common mistake: Assuming account exhaustion terminates all insurer obligations ignores a qualifying no-lapse guarantee and can omit its entire future claim exposure.
Official syllabus reference: [2]
50. UL emerging surplus from charges and spreads
UL profit can emerge from premium loads, expense charges, mortality charges, investment spread and surrender terms, offset by actual expenses, net claims and reserve requirements. Credited interest is a policy obligation; earned investment return is asset experience. Their difference creates a spread on the applicable balance, not necessarily on the opening account alone. Avoid counting an account deduction as profit while also ignoring its associated expense or insurance cost. A complete deterministic profit test reconciles assets, account movements, additional reserves and termination payments on a consistent timing basis.
Worked example: An account charge of 50 accompanies actual expense 60, producing expense margin -10 before interest effects. If 900 of assets earns 5% while the corresponding account earns 3%, the simple matching-balance investment margin is 18, partially offsetting that expense loss.
Common mistake: Equating all account deductions with profit overlooks the actual expenses and mortality costs those deductions are intended to finance.
Official syllabus reference: [2]
51. UL premium adequacy under interest and persistency stress
UL premium adequacy depends on account sustainability and insurer profitability, which are related but different tests. A premium can keep an account positive while failing a specified profit objective, or satisfy a guarantee's premium test while the account depletes. Stress earned returns, expenses, mortality and persistency while preserving contractual crediting floors and charges. Solve for premiums or reserves using the specified objective, then inspect duration-specific deficits. Persistency changes the period over which acquisition costs and guarantee losses can be recovered, so higher lapse is not uniformly favorable or unfavorable.
Worked example: A guarantee credits at least 2% while stressed assets earn 1%. Increasing the assumed crediting rate cannot repair that spread loss. A premium solve must project the negative spread, charges and expenses, then identify the premium meeting the chosen NPV or reserve objective.
Common mistake: Declaring a premium adequate solely because the projected account stays positive ignores insurer expenses, guarantee funding and the required profitability objective.
Official syllabus reference: [2]
Embedded Options in Life Insurance and Annuity Products
52. GMDB top-ups and death-time triggers
A guaranteed minimum death benefit commonly pays the larger of account value and a specified death guarantee when the insured dies during coverage. The insurer-funded top-up is max(G-F,0), distinct from the total death payment. Guarantee bases may be fixed, rolled up or ratcheted only if the contract specifies those features. Because death can occur at different times, valuation requires death-time probabilities and financial option values at those times. A maturity put alone does not reproduce a death-triggered benefit payable throughout the term.
Worked example: At death, account value is 72,000 and guaranteed death amount is 90,000. Total death payment is 90,000, composed of the account plus insurer top-up 18,000. If the account were 95,000, total payment would instead be 95,000 with no top-up.
Common mistake: Valuing the whole death payment as an option top-up double-counts the account component when account assets are already included elsewhere in the projection.
Official syllabus reference: [2]
53. GMMB decomposition into account and put
At a guaranteed maturity date, a minimum accumulation or maturity benefit can be written max(F_T,G)=F_T+max(G-F_T,0). The second term is a put-like top-up on the account. This decomposition separates the policyholder's fund from the insurer's guarantee exposure. Survival conditions, fees, withdrawals and guarantee-base changes can modify the valuation, so identify them before applying a plain European put formula. The payoff identity is exact at the stated date, but a pricing model adds assumptions about fund dynamics, discounting and financial market completeness.
Worked example: A maturity guarantee is 100 and terminal fund is 85. Total payment 100 decomposes into fund 85 and put payoff 15. If the fund ends at 120, the payment is 120 and the guarantee top-up is zero.
Common mistake: Using max(F_T-G,0) values a call-style excess above the guarantee rather than the shortfall the insurer must finance.
Official syllabus reference: [2]
54. GMIB income bases and annuity conversion
A guaranteed minimum income benefit protects an annuitization outcome, often by applying guaranteed conversion terms to a benefit base. Compare the guaranteed income with income purchasable from the actual account under the alternative terms. The base is typically a calculation quantity, not automatically a cash surrender amount. A capital-equivalent shortfall depends on the annuity price used at exercise, and may differ from the difference between base and account. Exercise conditions and policyholder behavior matter because the guarantee can be valuable only when the specified annuitization option is chosen.
Worked example: Benefit base is 100,000 and guaranteed income rate is 6%, giving annual income 6,000. Account 80,000 buys 5,000 at annuity price 16. The income enhancement is 1,000 annually, equivalent to capital 16,000 under that same price.
Common mistake: Treating an income benefit base as cash available for withdrawal confuses an annuitization calculation with a separately payable account balance.
Official syllabus reference: [2]
55. GMWB depletion and path-dependent guarantee costs
A guaranteed minimum withdrawal benefit promises specified withdrawals subject to contract conditions, potentially continuing after the account is exhausted. Withdrawals initially reduce the account, and insurer-funded payments emerge when account resources no longer cover guaranteed amounts. Investment returns, fees, withdrawal timing and benefit-base adjustments determine depletion, making the liability path-dependent. Total guaranteed withdrawals minus current account is not generally the option value. That subtraction works only as a simplified nominal shortfall under restrictive assumptions such as zero returns and fees with a fixed unconditional remaining payment stream.
Worked example: An account of 12,000 supports four guaranteed annual withdrawals of 5,000. With zero returns and fees, it finances 5,000, 5,000 and 2,000 of the third payment. The insurer supplies 3,000 then 5,000, totaling 8,000 before discounting.
Common mistake: Valuing GMWB as a single terminal put ignores the timing of withdrawals and the way earlier fund returns change later account exhaustion.
Official syllabus reference: [2]
56. Black-Scholes put valuation for maturity shortfalls
For a non-distributing fund under the standard Black-Scholes assumptions, a maturity shortfall put has value Ke^(-rT)N(-d2)-FN(-d1), where d1=[ln(F/K)+(r+σ²/2)T]/(σ√T) and d2=d1-σ√T. The model assumes continuous trading, constant parameters and lognormal fund dynamics. The rate r is continuously compounded in this expression. Fund fees or distributions require an adjusted formulation. Use the formula for the financial top-up, then incorporate any independent mortality condition separately rather than replacing volatility or discounting with mortality quantities.
Worked example: With F=K=100, r=0.04, σ=0.20 and T=1, d1=0.30 and d2=0.10. Using the corresponding normal probabilities gives put value about 6.004. That amount values the financial shortfall before applying any survival requirement or insurance expenses.
Common mistake: Substituting an annual effective interest rate directly for the continuous rate changes discounting and the d-values unless the rate is first converted consistently.
Official syllabus reference: [2]
57. Mortality weighting of financial guarantee values
When mortality is independent of financial markets and diversifiable under the valuation approach, a survival-contingent maturity top-up can be valued as survival probability times the corresponding financial option value. A death guarantee instead combines death probabilities or densities with option values at possible payment times. Dependence between mortality and markets breaks the simple product factorization. Survival weighting prices an expected pooled liability; it does not construct a complete financial hedge for one person's death event or remove systematic longevity risk. Keep the financial and insurance assumptions explicit.
Worked example: A one-year financial put costs 8 and the independent survival probability is 0.97. The expected pooled survival-contingent guarantee value is 7.76 under the stated approach. A death-contingent guarantee paid at varying dates needs separate timing weights rather than the same multiplier.
Common mistake: Multiplying by survival probability when the benefit is triggered by death applies the wrong insurance event and can reverse the intended weighting.
Official syllabus reference: [2]
58. Put delta and the replicating cash position
A non-distributing Black-Scholes put has fund delta N(d1)-1, which is negative. A portfolio matching the option liability holds that many fund units and cash B=P-ΔF. Its negative fund exposure reflects increasing guarantee cost when the account falls. The cash position offsets the short fund position so total portfolio value equals the positive option value. Delta changes with fund level, time and volatility, requiring rebalancing. Mortality-weighted expected liabilities can have scaled financial delta under appropriate independence assumptions, but separate mortality risk remains outside the financial replication.
Worked example: A guarantee put is worth 7 when fund value is 100 and delta is -0.35. A matching portfolio holds short 0.35 fund units and cash 42, since -35+42=7. The short position is an asset-side hedge of the put liability.
Common mistake: Holding positive put delta reverses the intended fund sensitivity and increases exposure to the falling-market scenario that makes the guarantee costly.
Official syllabus reference: [2]
59. Discrete hedging error, gamma and trading costs
Delta hedging matches local first-order fund sensitivity, but a finite interval leaves exposure to curvature, commonly summarized by gamma. The option delta changes as the fund moves, while the hedge remains fixed until the next trade. Transaction costs then reduce value each time the position is adjusted. More frequent trading can reduce discretization error while increasing costs, so continuous frictionless replication is a model benchmark rather than a literal costless operating strategy. Market gaps, parameter errors and basis mismatch add further risks beyond the simple discrete-time effect.
Worked example: A hedge must buy 0.20 fund units at price 100 to restore delta. The trade value is 20; at transaction cost 0.15%, cost is 0.03. The trade reallocates cash and fund, while the 0.03 directly reduces total hedge value.
Common mistake: Assuming a self-financing rebalance preserves value after transaction costs overlooks the cash consumed by trading and overstates realized hedge performance.
Official syllabus reference: [2]
60. Equity-linked profit tests with hedge financing
An equity-linked profit test should reconcile guarantee charges, account movements, benefit top-ups, hedge assets, hedge financing, transaction costs and reserves. A hedge gain offsets liability deterioration rather than automatically becoming distributable profit. Avoid counting both an option purchase cost and the same future guarantee payments without the corresponding hedge payoff. Stress fund returns, volatility assumptions, rebalancing frequency and policyholder behavior under explicit scenarios. Deterministic paths reveal cash-flow mechanics and funding strain, but they do not replace a stochastic valuation of nonlinear guarantees or establish a market-consistent option price.
Worked example: A year produces guarantee fees 4, top-up claims 12, hedge payoff 10 and trading costs 1. Before other expenses and reserve changes, net result is 4-12+10-1=1. Omitting the hedge payoff would misrepresent the financed guarantee strategy.
Common mistake: Recognizing hedge profits without the matching guarantee liability movement can report strong earnings precisely when the underlying insurance obligation has become more expensive.
Official syllabus reference: [2]
Official syllabus sources
- Advanced Long-Term Actuarial Mathematics (ALTAM) Exam
- October 2026 Syllabus with Learning Objectives/Outcomes and Readings