Why can two sound actuarial valuations of the same obligation differ?

They can answer different questions. A funding valuation estimates contributions under a funding policy; an accounting valuation measures amounts for financial reporting; an economic analysis may assess market-sensitive risk. Discount rates, demographic assumptions, margins and asset treatment follow the purpose and applicable framework. Reconcile the cash flows and assumptions before treating the difference as an error.

When should an actuary use experience rather than a manual assumption?

First check whether the experience measures the intended population, benefit, period and exposure. Adjust for changes such as trend, claim development and benefit design. Then assess its stability and credibility. A small group with volatile claims may call for a blend of adjusted experience and a relevant manual rate; a large, consistent group may support more weight on its own experience. The selected assumption should reflect the future risk being priced or valued.

How does risk transfer change an insurer's capital decision?

Reinsurance or a derivative can reduce a specified exposure, but the benefit depends on the contract's terms, basis risk, counterparty strength and cost. Compare capital and cash flows before and after the transaction, including collateral and possible recapture or hedge failure. Lower modeled risk does not by itself establish a better decision if the price of protection exceeds the value of the risk reduction.

FSA Practice Areas: An Applied Study Guide

Start with concept 1 Choose a topic

Fellowship work asks you to turn uncertain future cash flows into decisions that can be explained and defended. A reserve estimate depends on data development and assumptions; a product price also depends on selection, expenses, capital and customer behavior. The same figures may serve different purposes under statutory reporting, U.S. GAAP, IFRS or a funding valuation. This guide develops those distinctions through short, self-contained problems across the Society of Actuaries' seven FSA practice areas. Start with the purpose of each calculation, identify the risks and cash flows, choose a suitable method, and interpret the result for the decision maker. The topics survey course choices in the flexible pathway. Your four technical courses need not cover every area shown here.

Corporate Finance and ERM

1. Assign risk ownership and oversight

A sound ERM framework gives each risk an owner, a way to measure it, a limit, and a route for escalation. The business that creates a risk owns its day-to-day decisions and controls. An independent risk function helps define appetite, challenges assessments, monitors limits, and reports breaches across business units. Internal audit then tests whether the framework and controls work as designed, reporting independently to the board or its audit committee. These roles collaborate, but independence weakens if audit operates the control it later assesses. When assigning a proposed control, ask who can prevent the risk at its source, who should challenge that person's decisions, and who can provide assurance afterward. The CFE 101 study guide develops this governance work within the flexible FSA pathway.

Worked example: A sales unit proposes a new underwriting rule and records every exception. The risk team sets a hypothetical limit of 20 exceptions per month and monitors the count. If sales records 24, risk escalates the four-exception breach and asks for a response. Audit samples approved sales and tests whether exceptions were recorded correctly. Having audit approve each sale would compromise its later assurance.

Watch the distinction: Giving the risk team ownership of every underwriting decision sounds cautious, but it separates decisions from the people creating the exposure. Independent oversight should challenge and escalate business decisions.

Reference: [1]; [2]

2. Turn emerging risks into indicators

An emerging risk becomes manageable when a vague concern is translated into a measurable exposure, an indicator, and a response. A leading indicator changes before losses emerge, such as the share of critical systems awaiting a security patch. A lagging measure, such as paid cyber claims, confirms realized harm but may arrive too late to guide prevention. Define the indicator's denominator, observation date, data owner, and trigger so movement is interpretable. Compare it with an explicit risk appetite limit, then specify who investigates, what action follows, and when the result returns to decision makers. Examine possible false signals: a higher unpatched share might reflect newly discovered systems rather than deteriorating patch performance. That distinction changes the diagnosis, though both conditions may warrant attention.

Worked example: In a hypothetical cyber portfolio, the share of critical systems unpatched past the chosen deadline rises from 4% to 11%. Claims usually arrive months after an attack, so their current count is lagging evidence. Management's stated escalation trigger is 10%; 11% exceeds it by one percentage point. Escalate now, identify affected systems, prioritize remediation, and track whether the share falls.

Watch the distinction: Waiting for paid claims before acting confuses confirmation with warning. Claim reporting delays can conceal a growing exposure, while a well-defined operational indicator can prompt earlier investigation.

Reference: [2]

3. Read tail risk measures

Value at risk, or VaR, identifies a loss threshold at a stated confidence level and horizon. For a discrete distribution, find the smallest loss whose cumulative probability reaches that confidence level. Tail value at risk, or TVaR, averages losses in the worst fraction of outcomes. With a probability mass exactly at the cutoff, allocate only the portion needed to fill that tail; otherwise a simple conditional average can give the wrong answer. VaR is useful for setting a threshold, but says nothing about the size of losses beyond it. TVaR reflects their severity, although both depend on the modeled distribution and can miss omitted risks. State whether loss is positive, what time horizon applies, and whether results are gross or net of risk transfer before comparing portfolios.

Worked example: Assume one-year losses, in millions, are 0 with probability 90%, 10 with probability 5%, and 50 with probability 5%. Cumulative probability reaches 95% at 10, so 95% VaR is 10 million. The worst 5% consists solely of the 50-million outcome, making 95% TVaR 50 million. A larger final outcome could leave VaR unchanged while raising TVaR.

Watch the distinction: Treating 95% VaR as the largest plausible loss ignores the remaining 5% of outcomes. It marks a percentile, so compare tail severity separately.

Reference: [2]

4. Aggregate dependent risks

When two capital estimates represent risks measured on a compatible basis, a common correlation approximation combines them as √(C₁² + C₂² + 2ρC₁C₂). Correlation controls the modeled diversification: at perfect positive correlation, capital sums; with weaker dependence, the square-root result is smaller. This approximation assumes the chosen correlation reasonably describes joint behavior at the relevant tail and that the capital figures can be combined. Ordinary correlation may miss simultaneous extremes, so stress scenarios should test the benefit. Also distinguish statistical diversification from deployable capital. Group results may improve while capital remains trapped behind legal, regulatory, currency, or operational barriers. For a business decision, report the modeled reduction and assess whether resources can reach the entity that bears the loss when needed.

Worked example: Two subsidiaries have hypothetical standalone capital of 60 and 80 million, with correlation 0.25. Combined capital is √(60² + 80² + 2 × 0.25 × 60 × 80) = √12,400 = 111.36 million. The modeled benefit against the 140-million sum is 28.64 million. If one subsidiary cannot transfer funds promptly, some of that benefit may be unusable.

Watch the distinction: Subtracting the full modeled diversification benefit from each subsidiary's funding need counts the same benefit twice. Legal-entity capital needs and transfer limits still matter.

Reference: [2]

5. Trace a stress through the balance sheet

A balance-sheet stress should trace each shock through assets, liabilities, cash flows, and surplus. Begin with surplus equal to assets less liabilities. Revalue both sides under a consistent scenario, then add direct costs or cash demands that the valuations have not already included. For an interest-rate rise, fixed-income assets may lose market value, while the present value of fixed liabilities may fall. A linked lapse shock can change expected liability cash flows and force asset sales or surrender payments. This interaction is why adding separate one-factor sensitivities may misstate the result. Check the sign of every change and avoid counting surrender costs twice if they are already embedded in the liability remeasurement. A solvency assessment should also ask when cash is required, because a favorable present-value change does not necessarily supply immediate liquidity.

Worked example: Suppose a hypothetical combined rate and lapse stress reduces assets by 30 million and measured liabilities by 18 million. Separate surrender processing and liquidation costs, excluded from those values, add 7 million. Surplus changes by −30 − (−18) − 7 = −19 million. The liability decline offsets 18 million of asset loss, but the behavioral costs deepen the net reduction.

Watch the distinction: Calling an 18-million liability decline an 18-million surplus loss reverses the balance-sheet sign. Lower liabilities increase surplus, provided the related cash costs are included separately.

Reference: [2]

6. Choose a risk response

A risk response should be selected against both expected return and the organization's stated tolerance for adverse outcomes. Acceptance keeps the exposure and its expected profit; avoidance stops the activity; operational reduction changes its frequency or severity; and transfer shifts specified losses to a counterparty for a price. First define the risk measure and limit, then compare each option on the same basis. Include implementation cost, residual tail risk, timing, and new exposures such as reinsurer default or control failure. A cheaper response can be preferable if it meets the limit, but risk reduction need not be linear: a control may work in routine cases and fail during a systemic event. Document which assumptions make the recommendation hold, and monitor realized results after implementation.

Worked example: A hypothetical product earns 12 million in expected profit but has a 40-million tail loss against a 30-million limit. A control costs 2 million and reduces tail loss to 28 million; reinsurance costs 5 million and reduces it to 20 million. Both comply. Expected profit becomes 10 million with the control or 7 million with reinsurance, so choose the control if its effectiveness is credible.

Watch the distinction: Choosing the option with the smallest residual tail loss alone ignores its cost and the actual limit. The decision also requires a credible estimate of remaining risk.

Reference: [2]

7. Allocate capital to decisions

Standalone capital measures a unit in isolation. Marginal capital measures how much group capital changes when that unit is added or expanded, using the same risk model before and after the decision. Diversification can make marginal capital much lower than standalone capital, so return on marginal capital can better inform a group expansion choice. Match the numerator to the decision: use incremental expected earnings after costs, rather than the unit's entire historical profit, when evaluating an expansion. Also examine allocation stability, since marginal amounts can shift as the rest of the portfolio changes. A unit may look attractive because it offsets current group risks but lose that benefit under a common stress or after other units shrink. Capital held in a particular legal entity can constrain the decision even when group marginal capital is small.

Worked example: A hypothetical unit earns 9 million annually, needs 60 million of standalone capital, and increases group capital by 30 million. Standalone return is 9/60 = 15%; return on marginal capital is 9/30 = 30%. The second ratio better describes its contribution to this group's capital efficiency. Before expanding it, test whether the 30-million marginal estimate holds in stressed conditions.

Watch the distinction: Using standalone capital for every group decision discards diversification information. Using marginal capital without checking legal-entity funding and stressed dependence can overstate the practical benefit.

Reference: [2]

8. Value a corporate project

Project valuation starts with incremental free cash flows: cash available after operating costs, taxes, required investment, and working-capital changes. Discount each dated amount at a rate consistent with its risk and currency, then subtract the acquisition price and any costs omitted from the forecast. Synergies count only when they are incremental to the buyer and realistically achievable; overlapping revenue forecasts or savings already in the base business must be removed. A positive net present value indicates value at the stated hurdle rate, but financing capacity can still make the project impractical if near-term capital or liquidity requirements cannot be met. Use sensitivities around integration costs, timing, and persistency to identify what would reverse the recommendation. A transaction price should reflect these risks rather than merely the seller's projected earnings.

Worked example: A hypothetical acquisition costs 100 million and produces incremental free cash flow of 60 million at each of the next two year ends. At a 10% annual hurdle rate, present value is 60/1.10 + 60/1.10² = 104.13 million, giving NPV of 4.13 million. An omitted 8-million cost payable now changes NPV to −3.87 million; reject or renegotiate.

Watch the distinction: Treating projected synergy as value without deducting the cost of achieving it inflates NPV. Check whether cash flows, integration expense, and financing needs use consistent timing.

Reference: [2]

9. Challenge reported performance

Reported profit and cash flow answer different questions. Accrual earnings recognize revenue and expense under accounting rules, while cash flow reveals when money is collected and paid. To evaluate a business unit, reconcile earnings with changes in receivables, payables, and other working capital. Then test whether allocated headquarters costs reflect the unit's use of shared services. A revenue-based allocation may charge a fast-growing unit more even when its actual service use is stable, distorting its apparent margin and manager incentives. Reallocating cost can improve comparability, but it does not create group profit because the same shared expense remains elsewhere. Examine changes in accounting estimates, reserve releases, and one-time items as well. Reward a manager only after separating controllable operating performance from allocation conventions and weak collections.

Worked example: A hypothetical unit reports profit of 20 million after a 12-million headquarters charge based on revenue. A documented usage measure supports a 5-million charge, so comparable unit profit is 20 + 12 − 5 = 27 million. If receivables rise by 15 million with no offsetting working-capital changes, simplified cash conversion is 27 − 15 = 12 million. Investigate collections before praising the higher margin.

Watch the distinction: Calling the 7-million allocation adjustment new company profit is incorrect; it shifts cost between units. Higher reported earnings also need a cash-flow check.

Reference: [2]

Group and Health Insurance

10. Match health benefits to risk

Benefit design changes both who pays a claim and how members use care. Medical coverage often combines deductibles, copayments, coinsurance, networks, and out-of-pocket limits; less cost sharing can improve access while increasing utilization. Dental benefits may emphasize preventive visits and place different limits on costly restorative work. Pharmacy designs can steer members toward suitable lower-cost drugs through tiers or formularies, but switching rules may affect adherence. Group life pays on death rather than reimbursing a service, so its main pricing drivers include insured amounts, mortality, and eligibility rather than visit frequency. For each provision, distinguish allowed cost, insurer payment, member payment, and any longer-term health effect. A sponsor's decision should compare expected cost with access, employee value, selection, and administrative consequences. The GH 101 syllabus places benefit design and pricing together.

Worked example: An employer considers removing a hypothetical $20 office-visit copayment. Expected visits rise from 2.0 to 2.3 per member annually, with allowed cost of $100 per visit. Annual allowed cost rises from 2.0 × $100 = $200 to 2.3 × $100 = $230, an increase of $30 per member. The sponsor should also model the removed member payment and any later savings from timely care.

Watch the distinction: Equating a $20 copayment removal with a $20 increase in total medical cost misses utilization. Allowed cost, member payment, and sponsor payment must be modeled separately.

Reference: [3]

11. Build a medical manual rate

A manual rate estimates claim cost for a defined population and benefit design before converting that cost into premium. Start by aligning claim dates, covered exposure, benefit terms, and any pooling adjustments. Divide adjusted claims by member months to obtain a per-member-per-month base. Project from the experience midpoint to the rating period using a trend assumption suited to the claim category; medical, dental, pharmacy, and group life may have different drivers. Then add expenses, taxes where applicable, and target margin in the correct form. If those items are percentages of premium, divide projected claims by one minus their combined share. If they are fixed amounts per member, add them instead. Check that the resulting premium covers the projected claims and loads, and examine actual-to-expected results after the period begins.

Worked example: Hypothetical adjusted annual claims are $4.8 million over 20,000 member months, giving $4,800,000/20,000 = $240 per member month. One year of 8% trend produces $240 × 1.08 = $259.20. If expenses and margin together equal 15% of premium, indicated premium is $259.20/0.85 = $304.94 per member month, rounded to cents.

Watch the distinction: Multiplying claims by 1.15 would underprice a 15% premium-based load. At that price, the load is less than 15% of premium.

Reference: [3]

12. Blend group experience and manual cost

Experience rating blends a group's own claims with a broader manual estimate. Before blending, bring both to the same coverage period, benefit design, geography, demographic mix, and trend basis; otherwise the difference reflects inconsistent inputs rather than the group's risk. Credibility expresses how much weight the group's experience deserves, given exposure size and volatility. The complement goes to the manual estimate. Large claims deserve separate investigation because a rare, nonrecurring event can dominate a small group's experience, while excluding every costly claimant would systematically understate prospective cost. Consider whether the condition is likely to recur, how pooling or stop-loss applies, and whether the manual already includes comparable events. Finally, add expenses and margin if the blended figure represents claims rather than premium. The GH 101 guide provides focused practice on pricing decisions.

Worked example: A hypothetical group's experience and the manual have both been adjusted to the coming year's benefits and exposure. Claims estimates are $330 and $300 per member month, respectively, with 40% credibility on experience. The blend is 0.40 × $330 + 0.60 × $300 = $312. Review a one-time transplant claim and applicable pooling before using $312 as prospective claim cost; premium loads remain additional.

Watch the distinction: Blending raw paid claims with a trended manual rate mixes time and completion effects with risk. Align the estimates first, then apply credibility.

Reference: [3]

13. Select an employer funding approach

A fully insured arrangement replaces much claim volatility with a known premium, although renewal prices can still change. Under self-funding, the employer pays claims and administration directly, retaining both favorable experience and adverse deviations. Stop-loss coverage can cap specified large or aggregate losses, subject to contract terms and exclusions. Level-funded arrangements usually collect a stable monthly amount combining expected claims, administration, and protection, with settlement of favorable or adverse experience governed by the agreement. Compare arrangements using the same covered benefits, expected claims, fees, taxes, cash timing, and plausible tail outcomes. An employer with limited liquidity may rationally pay more in expected cost for predictability. Conversely, a large employer with diverse members and sufficient reserves may be able to retain risk efficiently. Examine who owns claims data and how run-out claims are handled when funding changes.

Worked example: A hypothetical employer expects $2 million in claims and $200,000 in administration, so self-funded expected cost is $2.2 million. A comparable fully insured premium is $2.4 million, a $200,000 expected difference. If a plausible adverse deviation is $600,000, self-funded cost could reach $2.8 million. An employer unable to absorb that outcome should favor insurance or price suitable stop-loss protection before deciding.

Watch the distinction: Comparing only expected self-funded claims with the insured premium omits administration and volatility. Funding choice also depends on the employer's ability to pay when claims arrive.

Reference: [3]

14. Estimate unpaid health claims

Health claims often develop after the month in which care occurs. A completion factor estimates the share of ultimate incurred claims already paid or reported at a valuation date. Divide observed claims by the factor to estimate ultimate claims, then subtract observed amounts for the unpaid or unreported balance on that basis. Be precise about the data: paid-to-ultimate factors produce total unpaid claims, including reported but unpaid amounts; reported-to-ultimate factors can help isolate incurred but not reported claims. Group claims by incurred month and development age, and inspect seasonality, benefit changes, large claims, and processing speed. A faster payment system raises the paid share at a given age; applying an older, lower factor would inflate the estimate. Compare recent cohorts and operational metrics before changing a selected factor.

Worked example: For a hypothetical incurred month, $720,000 has been paid and an applicable paid-to-ultimate factor is 80%. Estimated ultimate claims are $720,000/0.80 = $900,000, leaving $180,000 unpaid. That $180,000 includes any reported claims awaiting payment, so it is not automatically all IBNR. If processing has accelerated, test whether 80% remains representative.

Watch the distinction: Labeling ultimate less paid entirely as IBNR confuses unreported claims with reported claims still unpaid. The distinction depends on whether the starting data are paid or reported.

Reference: [3]

15. Interpret U.S. health insurer capital

For a U.S. health insurer, claim liabilities, accounting earnings, statutory surplus, and risk-based capital answer separate questions. A higher estimate of unpaid claims increases the liability and generally reduces current earnings and surplus before tax or other effects. Statutory and GAAP results may differ because their recognition and measurement rules serve different reporting purposes. Risk-based capital is a separate calculation of required capital for specified risks, including insurance and asset exposures; its components may respond indirectly when experience or business volume changes. The reserve itself is not an equal-sized capital requirement. To assess financial strength, compare available statutory capital with the applicable required-capital measure, while reviewing reserve adequacy, underwriting results, and cash needs. State the accounting basis and whether a change affects the numerator, denominator, or both before interpreting a ratio.

Worked example: A hypothetical U.S. insurer adds $10 million to unpaid claim liabilities after new experience. Assuming no tax or other entries, pre-tax earnings and statutory surplus each decline by $10 million. If required risk-based capital was $40 million, it does not mechanically become $50 million. Recalculate the requirement from its risk inputs and assess how the lower available surplus changes capital coverage.

Watch the distinction: Calling a reserve increase an identical increase in required capital conflates a booked liability with a risk formula. It can reduce available surplus without changing required capital dollar for dollar.

Reference: [3]

16. Read Canadian insurance results

Canadian insurance reporting and capital assessment use different lenses. Under IFRS 17, fulfillment cash flows reflect probability-weighted future cash flows, discounting, and a risk adjustment for non-financial risk. For profitable groups, a contractual service margin can defer unearned profit; the full contract measurement therefore requires more than a simple net outflow. Canada's Life Insurance Capital Adequacy Test, or LICAT, assesses available capital against prescribed risk requirements and aggregation rules. A contract can have a modest measured liability yet consume substantial capital because its downside is volatile. When reviewing results, first identify the contract group and measurement model, then reconcile liability components and earnings emergence. Separately examine available capital, required capital, and stresses. Do not interpret the IFRS 17 risk adjustment as the LICAT capital requirement: their objectives and calculations differ.

Worked example: A hypothetical Canadian contract group has discounted future outflows of 110 million, inflows of 100 million, and a 3-million non-financial risk adjustment. Its illustrative net fulfillment cash flow is an outflow of 110 − 100 + 3 = 13 million. Further IFRS 17 components depend on the group's measurement. The 13 million alone does not determine LICAT required capital.

Watch the distinction: Using the IFRS 17 risk adjustment as a proxy for LICAT required capital merges accounting measurement with a solvency assessment. Their inputs and aggregation rules serve different questions.

Reference: [3]

17. Evaluate provider payment and care programs

Provider payment changes incentives. Fee-for-service pays for each covered service, which can reward additional volume. An episode bundle pays a defined amount for specified care, encouraging coordination and cost control while exposing the provider to cost variation within the bundle. Contract evaluation must identify which services and complications are included, quality safeguards, risk adjustment, and who bears unusually costly cases. A care program should be assessed against an expected-cost comparison that reflects participant risk and underlying trend, not simply against participants' earlier spending. Calculate gross savings as adjusted expected cost less actual or projected cost, then deduct program expense. Check whether selection, regression to the mean, or coding changes explain part of the difference. A financially attractive program that worsens outcomes or access may still be unsuitable.

Worked example: A hypothetical provider receives $12,000 for a defined episode and incurs $11,100 in covered costs, leaving $900 before overhead and other obligations. Separately, a care program reduces risk-adjusted expected spending from $1,000 to $940 per participant. Gross savings are $60; after $25 program cost, net savings are $35 per participant. Review quality and selection before expansion.

Watch the distinction: Calling every decline in participant spending a program saving ignores what would have happened without the intervention. Use a comparable, risk-adjusted expected-cost control.

Reference: [3]

18. Normalize health risk before comparison

Health-status risk scores express predicted cost relative to a reference population. Dividing observed cost by a score can give a simple standardized cost for comparing groups with different measured morbidity, provided the scores use the same model, period, benefits, and calibration. That adjustment does not prove efficiency. More complete diagnosis coding can raise a score without changing actual health, making normalized cost look lower. Conversely, limited access can suppress recorded diagnoses and care, distorting both numerator and denominator. Social conditions such as housing, transport, and income can affect health and service use in ways a clinical score misses. Before attributing a standardized difference to provider performance, compare coding intensity, benefit design, provider prices, utilization mix, and member selection. Use uncertainty ranges when the groups are small.

Worked example: Hypothetical annual claims are $1,200 per member for Group A with score 1.20 and $1,050 for Group B with score 1.00. Simple normalized cost is $1,200/1.20 = $1,000 for A and $1,050/1.00 = $1,050 for B. A appears $50 lower after adjustment, but verify coding and access before treating that difference as efficiency.

Watch the distinction: A lower normalized cost does not by itself prove better management. Differences in diagnosis capture, prices, and unmet care can produce the same numerical result.

Reference: [3]

General Insurance

19. Prepare general insurance exposure

General insurance analyses need a consistent exposure base and premium rate level. Written exposure attaches when a policy is issued; earned exposure recognizes the portion of its coverage period that has passed. For a uniform 12-month policy, six covered months contribute half a policy-year. Written premium similarly becomes earned over coverage, subject to the policy's earning pattern. To compare historical losses with prospective premiums, restate earned premium to the selected current rate level, accounting for when rate changes applied and which policies they affected. Do not inflate a policy's actual earned premium merely because a later rate change occurred; that produces an on-level analytical amount for comparison. Check cancellations, endorsements, varying limits, and exposure definitions before calculating loss ratios or pure premiums. The GI 101 syllabus includes these preparation steps.

Worked example: A hypothetical 12-month policy begins July 1 with $1,200 written premium and uniform earning. By December 31 it contributes six months, or 0.5 policy-year, and $600 actual earned premium. If the same coverage would cost 10% more at the selected current rate level, its analytical on-level earned premium is $600 × 1.10 = $660. Actual booked earned premium remains $600.

Watch the distinction: Using written premium as the denominator for calendar-year losses overstates the premium tied to partial-year coverage. Separate actual earned premium from its on-level analytical restatement.

Reference: [4]

20. Select ultimate claim methods

Ultimate claim methods differ in how strongly they react to observed emergence. The development method divides reported claims by the estimated proportion reported, effectively assuming the current experience scales to ultimate. Bornhuetter-Ferguson adds reported claims to the expected ultimate amount associated with the unreported portion. It therefore relies more on an independent expectation when a period is immature. Both require a credible development pattern; the expected method also requires a sound exposure, rate, or loss-ratio basis. Investigate unusually large claims, changes in case reserve adequacy, settlement speed, and mix before selecting a result. As maturity rises, the unreported portion shrinks and Bornhuetter-Ferguson gives more weight to emerged claims. Compare methods and prior estimates, then explain why the selected ultimate is reasonable for that particular period rather than averaging results automatically.

Worked example: Suppose hypothetical reported claims are 600 thousand dollars at an estimated 60% maturity, with independent expected ultimate claims of 900 thousand dollars. Development gives 600/0.60 = 1,000 thousand dollars. Bornhuetter-Ferguson gives 600 + 900 × (1 − 0.60) = 960 thousand dollars. If an unusual early large claim drove reported experience, the expected-based 960 may be steadier, pending claim review.

Watch the distinction: Assuming the lower Bornhuetter-Ferguson result is always conservative reverses the logic. Depending on observed emergence and expected ultimate, either method can produce the higher estimate.

Reference: [4]

21. Correct distorted claim development

An incurred claim triangle combines payments and case reserves, so its development factors reflect both claim emergence and how adjusters estimate open claims. A sudden increase in case reserve adequacy raises current incurred values and can make later incurred age-to-age factors look smaller even when ultimate claim cost is unchanged. Faster settlement creates a different distortion: paid losses rise earlier, reducing later paid development. Diagnose the cause by comparing paid and incurred triangles, closed-claim counts, settlement rates and average case reserve per open claim by accident period. For a reserve adequacy change, restate older case estimates on a comparable basis before selecting factors. For a settlement change, adjust the paid triangle or use a method less dependent on the disrupted pattern. The GI 101 syllabus includes these development adjustments.

Worked example: Suppose old-style incurred claims at 12 months were $800,000 and developed by a factor of 1.50 to $1.20 million. After case reserves per open claim doubled, comparable claims show $1.00 million incurred at 12 months and a factor of 1.20, again producing $1.20 million ultimate. Applying 1.20 to the old $800,000 gives only $960,000, understating ultimate by $240,000.

Watch the distinction: A lower incurred development factor does not prove claims have become cheaper. First check whether case estimates, closure speed or claim mix changed before applying that factor to another period.

Reference: [4]

22. Separate frequency and severity trend

Pure premium is expected claim count per exposure multiplied by expected cost per claim. Project frequency and severity separately because they respond to different forces: underwriting and safety can affect claim counts, while repair costs, medical costs and litigation can affect claim size. For annual changes, multiply the factors rather than adding percentages; their product captures the interaction. Set the trend interval from the average accident date underlying experience to the average accident date in the prospective coverage period. Check whether exposure units have changed: a premium or insured value base that already grows with inflation can absorb part of an apparent severity increase. Also separate shifts in deductible, limits or risk mix from underlying trend. A useful GI 101 study guide can help connect these adjustments to a rate indication.

Worked example: Assume current claim cost is $500 per policy-year, frequency will fall 2% over one year, and severity will rise 6%. The projected pure premium is $500 × 0.98 × 1.06 = $519.40 per policy-year, up 3.88%. If the 6% severity assumption already includes repair inflation, adding another 6% total-loss trend would count that effect twice.

Watch the distinction: Adding frequency and severity trends gives 4%, which misses their interaction. The correct combined change is 3.88%; also check whether inflation has already entered the exposure measure or severity estimate.

Reference: [4]

23. Price deductibles and limits

A deductible assigns the first layer of a covered loss to the insured; a limit caps the insurer's payment. Specify whether the stated limit applies to ground-up loss or to payment after the deductible, because the formulas differ. If a ground-up loss is capped at limit L and a deductible d lies below L, the insurer pays min(X,L) minus min(X,d) for loss X. Expected insurer loss therefore equals expected limited loss at L minus expected limited loss at d. This decomposition lets you price options using limited-loss estimates or a loss distribution. Then add claim handling expense, other expenses, risk margin and target profit according to the pricing basis. Check that both limited-loss values use the same coverage terms, exposure period and loss distribution. Higher limits can also have different development and severity trend than primary layers.

Worked example: For a hypothetical policy with a $5,000 ground-up limit and $500 deductible, expected loss capped at $5,000 is $850 per exposure. Expected loss capped at $500 is $300. Expected insurer payment is $850 − $300 = $550 per exposure. If fixed expenses are $50 and the desired risk and profit load is $40, an illustrative indicated price is $640.

Watch the distinction: Subtracting the $500 deductible from the $850 expected limited loss is wrong: many losses are smaller than $500. Subtract expected retained loss, here $300, rather than the contractual deductible amount.

Reference: [4]

24. Price a reinsurance layer

An occurrence excess treaty pays the portion of each covered event above its attachment point, subject to the layer width. For a $1 million layer excess of $2 million, ceded loss is min(max(X − $2 million, 0), $1 million) for event loss X. Apply this formula to each outcome before weighting by probability; averaging ground-up losses first loses the nonlinear effect of attachment and exhaustion. The resulting expected ceded loss is only the starting point for a quote. The reinsurer also needs to cover expenses, uncertainty, capital use and contract features such as reinstatement premium or aggregate limits. Check whether the stated probabilities represent mutually exclusive annual outcomes or separate events that could occur together. A cedant should compare the quote with the reduction in its own tail exposure and capital needs.

Worked example: Assume mutually exclusive annual losses of $1 million, $2.5 million and $4 million occur with probabilities 80%, 15% and 5%. The layer pays $0, $0.5 million and $1 million respectively. Expected ceded loss is 0.80 × $0 + 0.15 × $500,000 + 0.05 × $1,000,000 = $125,000. A $140,000 premium leaves $15,000 before reinsurer expenses and risk compensation.

Watch the distinction: The $15,000 difference between premium and expected loss is not pure profit. Administration, adverse outcomes, capital costs and contract provisions can consume it, so assess the whole loss distribution.

Reference: [4]

25. Measure reserve uncertainty

A central unpaid-claim estimate represents a selected expected outcome, while reserve uncertainty describes how actual payments may differ. Process risk concerns random future claim outcomes even if model parameters are correct. Parameter risk reflects uncertainty in estimated development patterns or other inputs. Model risk includes a faulty structure or a missing exposure and may not be captured by either standard deviation. When process and parameter errors are independent, their variances add; if correlated, include the covariance term. Select a risk margin using the desired decision criterion, such as a percentile or capital measure, rather than treating one standard deviation as universally sufficient. Catastrophe output needs separate attention because a rare, highly dependent event can dominate tail loss while barely affecting ordinary development history. Reconcile the uncertainty analysis with portfolio concentration, reinsurance and recent changes in claim handling.

Worked example: A hypothetical portfolio has central unpaid claims of $50 million, process standard deviation $4 million and parameter standard deviation $3 million. Assuming independence, combined variance is 4² + 3² = 25 million dollars squared, so standard deviation is $5 million. A $5 million addition produces $55 million, but it is only one modeled standard deviation above the central estimate and does not cover every catastrophe.

Watch the distinction: A reserve margin equal to one standard deviation is not a guaranteed upper bound. Tail shape, dependence, parameter error and omitted catastrophe exposures can all make larger adverse outcomes plausible.

Reference: [4]

26. Connect legal change to claim cost

Liability claim severity can rise faster than general prices when litigation behavior, legal interpretation, damage awards or settlement practices change. This excess is often called social inflation, but the observed difference also may reflect a changing mix of claim types or policy limits. To diagnose it, compare like-for-like claims at a common maturity, coverage definition and limit; separate economic inflation in wages or medical costs from changes in legal outcomes. Higher awards affect insurer cost only to the extent that policies cover them and limits are available. They can also lengthen settlement, increasing defense expense and delaying emergence in incurred triangles. In a pricing projection, avoid treating a short run of unusually large cases as a permanent trend without examining frequency, severity distribution and case reserve adequacy. In reserving, test whether older accident periods need a different development pattern.

Worked example: Suppose average bodily injury settlement rises from $100,000 to $125,000, while a broad price index rises 5% over the same period. Nominal severity rises 25%. The increase beyond general inflation is 1.25/1.05 − 1 = 19.05% on a multiplicative basis. Before projecting 19.05%, compare claim mix, available limits, legal venues and settlement speed across the two periods.

Watch the distinction: Calling the whole 25% increase social inflation ignores ordinary economic inflation and changes in claim mix. A defensible selection separates those effects before projecting future covered severity.

Reference: [4]

27. Read a U.S. insurer's reserve opinion

For a U.S. property and casualty insurer, the booked unpaid loss and loss adjustment expense amount is a financial statement figure; the actuary develops an independent estimate or reasonable range to assess it. Reconcile both on the same gross or net basis, valuation date, coverage and expense scope. Then compare the booked amount with the supported estimate, investigate material differences, and consider subsequent information and uncertainty. Materiality concerns the effect a misstatement could have on users of the statement, including its relation to surplus and other financial measures; a fixed dollar difference has no universal answer. The signing actuary's opinion and supporting report require a reasoned conclusion and clear communication of relevant qualifications or limitations. The July 2027 GI 302 syllabus covers U.S. statutory reporting and signing responsibilities; other general insurance courses discuss opinions in broader jurisdictions.

Worked example: A hypothetical U.S. insurer books $82 million of unpaid losses and loss adjustment expenses. After reconciling the same net basis, the actuary supports $85 million to $95 million. The booked figure is $3 million below the lower bound. If surplus is $30 million, that difference equals 10% of surplus; investigate data and assumptions and assess materiality before stating an opinion.

Watch the distinction: A selected range does not automatically make every point outside it a prescribed opinion category. Reconcile definitions, assess materiality and document why the booked amount differs before reaching that conclusion.

Reference: [4]

Individual Life and Annuities

28. Map life and annuity guarantees

Term insurance pays on death during a specified coverage period, concentrating mortality risk and often ending without a payment. Whole life extends death coverage and commonly builds cash value, so investment performance and lapse behavior matter alongside mortality. Universal life separates account value, charges and credited interest; guarantees may limit how much adverse investment or lapse experience can be passed to policyholders. A payout annuity reverses the survival exposure by continuing payments while the annuitant lives, creating longevity risk. Identify who receives cash and when, then ask which assumptions increase the present value of benefits. Mortality and longevity exposures can offset, but the offset depends on ages, underwriting, benefit sizes and common shocks. Product design also differs from group insurance benefits, whose coverage and pricing structure may change with the sponsoring group.

Worked example: Suppose a ten-year term policy pays $100,000 on death during its term, while a payout annuity pays $10,000 at each year-end the annuitant survives. Earlier death makes a term claim more likely but removes later annuity payments. If the annuitant survives three years, the annuity pays $30,000 before discounting; no term death payment follows solely from that survival outcome.

Watch the distinction: Owning both life policies and annuities does not guarantee a mortality hedge. Different insured populations, ages, contract terms and catastrophe exposure may prevent their cash flows from offsetting when needed.

Reference: [5]

29. Test life product profitability

Product profitability should be measured through cash available to shareholders after claims, expenses, taxes, changes in reserves and required capital. Capital posted at issue is a shareholder outflow even if it may later be released; the timing of that release affects value and internal rate of return. Discount distributable earnings at a rate consistent with the chosen shareholder hurdle and compare their present value with initial investment. Then test whether expected earnings rely on optimistic lapse, mortality or investment assumptions. Value of new business, return on equity and internal rate of return answer related but distinct questions, so state which decision criterion is being used. An apparently positive margin may disappear when acquisition costs or higher capital requirements are included. For a price recommendation, also examine sensitivity and whether the product remains attractive under plausible adverse experience.

Worked example: Assume a hypothetical policy requires $100 of shareholder capital at issue and distributes $60 at each of the next two year-ends, including releases of that capital. At a 10% hurdle, present value is $60/1.10 + $60/1.10² = $104.13, so value added is $4.13. An omitted $6 acquisition expense at issue changes value to negative $1.87.

Watch the distinction: A product can report positive accounting profit while destroying shareholder value. Ignoring capital tied up at issue or discounting late releases too lightly overstates its attractiveness.

Reference: [5]

30. Set behavior assumptions from experience

Experience rates require a numerator consistent with the exposure denominator. For annual lapse, count terminations during observation and divide by policy-years exposed to possible lapse, with appropriate treatment of partial years and competing exits such as death. Segment the study by duration, age, product, premium pattern and other factors that may change behavior; use credibility when a segment is sparse. A premium jump can cause selective lapse if healthier policyholders have better replacement options, leaving a less healthy in-force pool. Conversion options can have the opposite selection problem when policyholders with worse expected mortality exercise them. Benefit utilization, such as an annuity guarantee election, may respond to market conditions and should not be projected from an unconditional average. Document whether the assumption is for pricing, reserving or another purpose because the decision may require different margins or scenario treatment.

Worked example: A hypothetical term block has 800 lapses over 10,000 policy-years at risk, giving an observed annual rate of 800/10,000 = 8%. After a premium increase, suppose healthy lives lapse at 12% and impaired lives at 4%. A uniform 8% projection would misstate the retained mix: more impaired lives persist, increasing expected death cost relative to an unsegmented projection.

Watch the distinction: The observed 8% lapse rate is not automatically suitable for every future year. Exposure mix, duration and premium changes can alter both lapse frequency and the mortality of those who remain.

Reference: [5]

31. Manage non-guaranteed benefits

A guaranteed benefit is a contractual minimum; an adjustable crediting rate or participating dividend can change within the product's contractual and governing constraints. For an account-based policy, investment yield less credited interest contributes to spread earnings, but that spread also supports expenses, guarantees and capital. Lowering a non-guaranteed rate may raise immediate spread and simultaneously increase lapses, surrenders or complaints. The right analysis projects both sides: direct interest savings and changes in future premium, fee, surrender and claim cash flows. Segment policyholders by guarantee level and sensitivity to competitors' rates; a change that affects only balances above a guarantee has a different effect from one constrained by a minimum. Test adverse investment scenarios and explain the change clearly to stakeholders. A policyholder dividend likewise depends on the relevant experience and dividend approach, rather than behaving like a fixed debt coupon.

Worked example: A hypothetical account block has $10 million of balances, earns 5% and credits 4%. Annual gross spread is $10 million × (5% − 4%) = $100,000. Reducing crediting to 3.5% raises initial spread to $150,000, a $50,000 gain. If resulting lapses remove $2 million of balances, the new spread becomes $8 million × 1.5% = $120,000 before other effects.

Watch the distinction: The full $50,000 first-year interest saving is not necessarily a durable profit increase. Lower balances, surrender costs and future guarantee exposure can change the result.

Reference: [5]

32. Build a life reserve projection

A life reserve projection places future premiums, benefits and expenses on a consistent timeline and discounts them under the stated valuation basis. For each scenario, account for mortality, lapse, policyholder options and investment returns where relevant. Net liability cash flow is the present value of future outgo less future income, subject to the measurement rules of the jurisdiction and product. U.S. principle-based valuation can use modeled outcomes and prescribed components, while formulaic minimums or floors may still matter for a given contract. A simple best-estimate present value is therefore an illustration of the mechanism, not a complete statutory reserve. Check whether premium is due now or in the future, whether expenses include acquisition or maintenance costs, and whether reinsurance is reflected gross or net. Sensitivity to policyholder behavior is especially important where guarantees become valuable under stress.

Worked example: For a hypothetical one-year block, collect premium of 100 units now and pay expected benefits of 105 plus expenses of 5 units at year-end. At a stated 5% annual discount rate, projected net present outgo is (105 + 5)/1.05 − 100 = 4.76 units. That modeled amount would need comparison with any applicable components or floor before a statutory reserve is selected.

Watch the distinction: The 4.76-unit present value is not automatically the required reserve. A valuation framework may prescribe margins, scenarios, floors or separate components that change the reported liability.

Reference: [5]

33. Separate U.S. GAAP liability components

U.S. GAAP separates several accounting questions that a single cash-flow model may inform. The liability for future policy benefits measures obligations from eligible insurance contracts using the applicable measurement model. Deferred acquisition costs concern qualifying costs of obtaining business and their recognition over future coverage; they are an asset or separate balance, not a direct subtraction from future claims. A market risk benefit captures certain guarantees that expose an insurer to capital market risk and is measured separately where the classification applies. Product classification matters: traditional life, universal life and variable annuities do not all use identical liability mechanics. Trace each modeled cash flow to the appropriate component, then reconcile changes through earnings and other applicable reporting lines. The ILA 101 study guide provides broader product context for these distinctions.

Worked example: A hypothetical insurer pays $12 million of qualifying acquisition costs for a new block and estimates $80 million of future policy benefit liability. Recording a $12 million deferred acquisition cost balance does not turn the claim liability into $68 million; the two balances answer different questions. If a variable annuity guarantee meets market risk benefit criteria, that feature also requires separate measurement.

Watch the distinction: Netting acquisition spending against the claim liability conceals two different measurements. It also obscures how changes in future benefit assumptions and acquisition cost recognition affect reported results.

Reference: [5]

34. Interpret IFRS 17 service margin

Under IFRS 17, fulfillment cash flows combine expected future cash flows, discounting and a risk adjustment for non-financial risk. For a profitable group measured under the general model, the contractual service margin represents unearned profit at initial recognition, avoiding an immediate gain. As insurance service is provided, the margin is released using coverage units that reflect the quantity of benefits and expected coverage period. Changes in estimates relating to future service can adjust the margin, while other changes may affect current profit or loss under the applicable rules. An onerous group has no positive margin to absorb an expected loss at inception. Start by determining contract grouping and measurement model, then identify inflows, outflows, risk adjustment and timing before discussing earnings emergence. A simple arithmetic residual illustrates the margin but leaves out discount-rate effects and detailed subsequent measurement.

Worked example: Assume a hypothetical profitable group has present-value premium inflows of 120 units, future outflows of 100 and a non-financial risk adjustment of 5 at inception, with no other components. Net expected gain before the service margin is 120 − 100 − 5 = 15 units. Under a general-model illustration, a 15-unit contractual service margin defers that gain for recognition as coverage is provided.

Watch the distinction: The 15-unit expected gain is not immediate day-one insurance profit under this illustration. Treating the contractual service margin as spare cash also confuses an accounting liability component with liquidity.

Reference: [5]

35. Distinguish capital measures

Available capital is the eligible financial resource measured under a particular framework; required regulatory capital is the amount that framework calculates for specified risks. Their ratio indicates regulatory coverage, but its meaning depends on the jurisdiction's definitions and adjustments. Internal economic capital asks a different question, often estimating capital needed for a chosen tail-risk measure and time horizon. Compare them only after aligning valuation basis, risk coverage and fungibility across legal entities. A group may show ample aggregate capital while a subsidiary lacks transferable resources, and holding capital above a requirement does not by itself establish dividend capacity. Examine stress losses, liquidity, future new business, rating objectives and restrictions on distributions. The FSA pathway permits course choices across practice areas; capital methods in life and health should still be applied within their own jurisdictional frameworks.

Worked example: A hypothetical U.S. life insurer has 150 units of available regulatory capital, 100 units of required regulatory capital and a separately estimated internal economic capital need of 130 units. Coverage is 150/100 = 150%, while available capital exceeds the internal estimate by 20 units. Neither the 50-unit regulatory excess nor the 20-unit internal excess alone establishes a safe dividend amount.

Watch the distinction: Subtracting required capital from available capital does not automatically produce distributable cash. Capital definitions, stress exposure, trapped resources and liquidity needs may constrain any payment.

Reference: [1]; [5]

36. Evaluate life reinsurance cash flows

Yearly renewable term reinsurance typically transfers a defined share of death risk in exchange for recurring risk premiums, leaving many other policy cash flows with the cedant. Coinsurance transfers an agreed share of premiums, benefits and often reserve economics, making its cash-flow and accounting effects broader. For either arrangement, map gross policyholder obligations first, then ceded cash flows and recoverables. Reinsurance does not usually remove the direct insurer's obligation to its policyholder, so reinsurer default creates counterparty exposure. Assess collateral, credit quality, settlement timing and treaty provisions such as recapture or exclusions. Capital relief depends on genuine risk transfer and the applicable regulatory treatment, not merely on labeling a treaty as reinsurance. A proportional share of a claim is straightforward to calculate, but profitability also depends on allowances, future premiums, reserve movements and adverse-deviation sharing.

Worked example: A hypothetical insurer cedes 40% of a $1 million death benefit under proportional coinsurance. The reinsurer's contractual claim share is $1 million × 40% = $400,000, and the cedant's net share is $600,000 if recovery is made. If the reinsurer fails to pay, the insurer still owes the policyholder $1 million and must address the $400,000 recoverable.

Watch the distinction: A 40% cession does not mean the policyholder must seek 40% of the benefit from the reinsurer. The direct insurer remains responsible for the full contractual payment.

Reference: [5]

Investment

37. Inspect fixed-income cash-flow risk

Fixed-income securities may promise similar yields while producing very different cash flows. A corporate bond exposes the investor to issuer default and spread changes; a fixed-rate noncallable bond also loses market value when yields rise. A leveraged loan commonly has a floating rate, reducing some rate duration but adding borrower credit risk, covenant and liquidity concerns. Mortgage-backed securities pass through principal as borrowers repay mortgages, so falling rates can accelerate refinancing and return principal just when attractive reinvestment yields disappear. Rising rates can slow repayment and lengthen effective duration. Compare securities by projected cash-flow timing under rate and credit scenarios rather than headline yield alone. For an insurer, this matters because unexpected principal return or extension can worsen matching to liability payments. The INV 101 syllabus identifies these asset classes within portfolio management.

Worked example: Suppose a hypothetical mortgage pool yields 5% and $1 million of principal is expected to remain invested for another year. If lower rates prompt full refinancing and proceeds can earn only 3%, annual interest on that amount falls from $50,000 to $30,000, a $20,000 reduction. A noncallable corporate bond avoids this borrower prepayment mechanism but retains rate and credit risks.

Watch the distinction: A higher mortgage-backed security yield is not free extra income. Prepayment and extension alter cash-flow timing, so compare scenario returns and liability matching before preferring it to a bond.

Reference: [6]

38. Value illiquid alternatives

Private debt provides contractual interest and repayment but can be concentrated and hard to sell. Private equity depends on uncertain operating gains and an eventual exit; leverage can magnify both outcomes. Real estate generates property cash flow plus a sale value, with vacancy, maintenance, financing and appraisal uncertainty. Compare alternatives using dated net cash flows, fees and a common valuation basis, then test how long capital may be locked up. Appraised values may move slowly even when sale prices change, making reported return volatility look artificially low. A high projected internal rate of return can also hide poor outcomes if the exit is delayed or proceeds require further investment. For an insurer or pension investor, the decision depends on liabilities and cash needs as well as expected return; an asset that cannot be sold promptly may force costly borrowing during stress.

Worked example: A hypothetical property costs $100, pays $6 net rent at year-end and sells then for $105. Its one-year holding-period return is ($6 + $105 − $100)/$100 = 11% before fees and tax. If sale instead occurs at the end of year two for the same $105 with no second-year rent and $4 financing cost, net profit is $7 over two years, not an 11% annual outcome.

Watch the distinction: An 11% one-year property return should not be reused after the sale date changes. Delayed proceeds, financing costs and uncertain exit value alter both total and annualized returns.

Reference: [6]

39. Turn liabilities into an investment policy

An institutional investment policy starts with the liabilities and the investor's ability to bear risk. Estimate benefit or claim payments by time period, then set required liquidity for those payments and collateral calls. State the return objective, funded-status or surplus risk tolerance, investment horizon, allowable assets, concentration limits, rebalancing authority and oversight responsibilities. Only then choose an allocation. A high expected return is of little use if the investor must sell illiquid assets at a discount to meet near-term payments. Distinguish market risk from liquidity risk: a portfolio can be valuable on paper and still fail to produce cash when needed. Consider contributions, operating cash flow or a committed liquidity facility as possible sources, but test their reliability under stress. Governance should specify who may change the allocation and how departures from policy are monitored.

Worked example: A hypothetical pension plan holds $100 million and must pay $20 million next year. A proposal puts $95 million in illiquid assets and $5 million in cash, with no contributions or other inflows assumed. The near-term funding gap is $20 million − $5 million = $15 million. Reallocate at least $15 million to dependable liquid assets or establish a credible liquidity source before adopting the portfolio.

Watch the distinction: The $100 million total asset value does not mean the plan can pay $20 million next year. Payment capacity depends on when assets can produce cash without unacceptable loss.

Reference: [6]

40. Measure bond rate exposure

Modified duration approximates the percentage price change for a small parallel change in yield: percentage change is approximately negative duration times yield change. Multiply by market value to estimate the dollar effect. This linear estimate omits convexity, so accuracy weakens for larger moves or securities with embedded options. Overall duration also compresses a curve into one number. Key-rate sensitivities instead measure the response to a local movement at a selected maturity while holding other curve points approximately fixed. Two portfolios can have the same total duration yet react differently to steepening or flattening if their cash flows sit at different maturities. For asset-liability management, compare asset and liability key-rate dollar sensitivities with consistent signs and shock definitions, then hedge the gaps. Consider credit spreads, prepayment and basis risk separately, because Treasury rate hedges do not remove every source of bond value change.

Worked example: A hypothetical bond portfolio is worth $10 million and has modified duration 5. For a small parallel 1 percentage-point yield rise, estimated value change is −5 × 0.01 × $10 million = −$500,000, leaving about $9.5 million before convexity. If its liability is sensitive mainly to 20-year yields, a hedge matching total duration but concentrated at five years may still leave a large curve-shape exposure.

Watch the distinction: Matching asset and liability duration does not ensure an interest-rate hedge. Their key-rate profiles can differ, so a nonparallel yield-curve move may change surplus substantially.

Reference: [6]

41. Attribute portfolio performance

Start with a return measured over the same period, in the same currency, and after a consistent treatment of fees. Active return is portfolio return minus benchmark return. Allocation effect asks whether the manager overweighted segments that performed well in the benchmark. Selection effect asks whether holdings beat their segment benchmarks. One consistent method calculates allocation from portfolio weight minus benchmark weight, multiplied by segment benchmark return; selection uses portfolio weight multiplied by portfolio segment return minus segment benchmark return. These effects sum to active return when weights cover the whole portfolio and there are no separate interaction terms. A useful benchmark must reflect the mandate’s permitted assets and risk exposures. Otherwise, an apparent selection gain may reward taking risks the benchmark never represented. The INV 101 syllabus includes benchmark selection and performance attribution.

Worked example: Suppose the benchmark holds 50% each in sectors A and B, which return 7% and 5%. Its return is 6%. The portfolio holds 75% in A, returning 9%, and 25% in B, returning 5%; its return is 8%. Allocation adds (0.75 − 0.50) × 7% + (0.25 − 0.50) × 5% = 0.5 percentage point. Selection adds 0.75 × (9% − 7%) = 1.5 points. Together they explain the 2-point excess, provided this benchmark suits the mandate.

Watch the distinction: A 2-point excess return does not establish manager skill. Check the benchmark and separate favorable sector positioning from returns earned within each sector.

Reference: [6]

42. Estimate expected credit loss

For a fixed exposure over a stated horizon, expected credit loss is exposure at default multiplied by probability of default and loss given default. Loss given default is the fraction unpaid after recoveries, so recovery timing and collateral values matter. Expected loss is an average allowance for credit cost, not the worst plausible outcome. A portfolio assessment also needs default dependence: borrowers sharing an industry, sponsor, guarantor, or economic driver may fail together. Ten individually small positions can therefore create one large tail exposure. Separate the expected loss estimate from capital or limits intended to absorb unexpected loss. When comparing loans, align the horizon and whether exposure can grow through undrawn commitments. Test whether default probabilities and recovery assumptions remain credible under stress; weaker recoveries often coincide with more defaults.

Worked example: Assume a $10 million loan has a 2% one-year default probability, 40% loss given default, and unchanged exposure through the year. Expected loss is $10 million × 0.02 × 0.40 = $80,000. Ten identical loans produce $800,000 of expected loss by addition. If all borrowers belong to one group, a shared failure can affect several loans together; the $800,000 mean remains, but treating defaults as independent understates tail risk.

Watch the distinction: Dividing exposure among ten loans does not automatically diversify credit risk. Shared ownership or economic drivers can make simultaneous defaults far more likely.

Reference: [6]

43. Price by replication

A European call plus a risk-free bond paying the strike at expiry has the same expiry payoff as one share plus a European put, assuming identical strike, expiry, underlying, and settlement terms. Both combinations pay the greater of the stock price and strike. Their current prices must therefore agree in a frictionless market: call price plus present value of strike equals stock price plus put price. This is a replication argument; it does not require a forecast of the stock’s expected return. A price gap suggests buying the cheaper combination and selling the dearer one, locking in the difference if both payoffs can be traded as assumed. Before declaring arbitrage, check dividends, exercise style, borrow constraints, financing rates, taxes, bid-ask spreads, and contract specifications. Any of these can alter the comparison or consume the apparent profit.

Worked example: For a hypothetical non-dividend stock, spot is $100, the strike’s present value is $95, and a matching European call costs $12. Parity gives put price $12 + $95 − $100 = $7. If the put is quoted at $9, stock plus put costs $109 while call plus strike bond costs $107. Sell the former and buy the latter for $2 now; their expiry payoffs match, before trading and financing frictions.

Watch the distinction: A put priced above parity is not automatically a usable $2 profit. Mismatched exercise rights, dividends, or trading costs can explain the quote.

Reference: [6]

44. Test a dynamic hedge

Delta measures the local change in an option’s value for a small change in its underlying price. To hedge a liability that gains value as the underlying rises, hold enough positive underlying exposure to offset that local sensitivity. The calculation is a starting position, not a permanent hedge: gamma changes delta as markets move, while volatility and interest rates change other parts of the option value. Rebalancing only at intervals leaves gains and losses between trades, and frequent trading adds costs. Insurance guarantees add another moving quantity, the number of policies remaining in force. Lapses, benefit use, and mortality can alter exposure when markets are stressed. Evaluate a proposed hedge against the risks it targets, its rebalancing rules, liquidity and collateral needs, and plausible behavior scenarios. Report residual risk rather than calling a delta-matched position fully hedged.

Worked example: Suppose an insurer owes 1,000 identical call-like guarantees, each with delta 0.60 underlying unit. The liability’s aggregate delta is 1,000 × 0.60 = 600 units, so an initial hedge holds 600 units of positive exposure. If a market fall changes each delta to 0.40 while 100 contracts lapse, the remaining liability delta becomes 900 × 0.40 = 360 units. The original hedge is then 240 units too large, before considering other risks.

Watch the distinction: Matching today’s delta does not lock in future hedge results. Delta and policy count change, and discrete trades leave exposure between rebalancing dates.

Reference: [6]

Retirement Benefits

45. Choose a retirement plan design

A defined benefit plan promises a benefit determined by a formula, often linked to pay and service. The sponsor must fund the promise as investment returns, salary growth, retirement timing, and longevity unfold. A defined contribution plan promises contributions to an account; the member generally bears investment and drawdown risk, though the sponsor still has design and administration duties. Hybrid designs divide risks in different ways. A cash balance design may express accruals as an account while promising interest credits; a target benefit design may aim for a lifetime pension while allowing specified adjustments. Identify the actual guarantee rather than relying on the plan’s label. Then compare sponsor budget stability, member income security, portability, workforce incentives, investment choices, and governance. A sponsor seeking predictable annual outlays may favor contributions, but member adequacy may require contribution levels and retirement-income options to be assessed separately.

Worked example: A hypothetical employer compares a $30,000 annual lifetime pension with a $10,000 annual account contribution. If a retiree lives 20 years after payments begin, the pension pays $600,000 undiscounted; if they live 30 years, it pays $900,000. The contribution promise remains $10,000 for each eligible contribution year, while the member’s eventual income depends on returns and withdrawals. Those amounts are not present-value equivalents; the comparison shows which party bears longevity risk.

Watch the distinction: An account-style description does not prove the sponsor has transferred all risk. Guaranteed credits or lifetime payment options can create obligations beyond annual contributions.

Reference: [7]

46. Measure retirement income adequacy

Replacement income compares retirement income with a chosen measure of working income, often final pay. Add pension, social insurance, and other recurring sources on a consistent gross or net basis, then divide by that pay measure. The ratio is a starting diagnostic, not a universal adequacy rule. Working households may save for retirement and pay employment costs that cease later, while retirees may face different taxes, housing costs, health expenses, and family support obligations. Timing also matters: a level pension loses purchasing power with inflation, and a temporary benefit can make the first year look stronger than later years. Build a spending projection by year, identify guaranteed and variable income, and test longevity and adverse investment outcomes. Explain whether any apparent shortfall is immediate, emerges later, or depends on discretionary spending assumptions.

Worked example: Suppose final pay is $80,000 a year, pension income is $28,000, and social insurance is $20,000. Total income is $48,000, so replacement is $48,000/$80,000 = 60%. Against estimated first-year spending of $45,000, the initial margin is $3,000. If the pension is level while spending rises, that margin may shrink; the 60% figure alone cannot determine long-term adequacy.

Watch the distinction: A replacement ratio below a familiar target does not by itself show inadequacy. Compare income with the retiree’s spending and how both change over time.

Reference: [7]

47. Price embedded retirement options

An early-retirement option changes both the benefit amount and the number of payments. An actuarially neutral adjustment would equate the expected present value of starting early with that of starting at normal retirement, using assumptions about survival, discounting, and payment form. A smaller annual pension can still cost more if payments begin several years sooner. Optional forms, such as a joint-life pension or guaranteed period, similarly change expected payments and may contain subsidies if conversion factors do not reflect their value. To assess sponsor cost, compare complete payment streams at the same valuation date; include commencement probabilities and whether employees who benefit most are more likely to elect the option. Also examine workforce effects: an early-retirement subsidy may achieve staffing goals, but it can concentrate elections and create a cash-flow strain.

Worked example: Assume certain survival, no discounting, and payments at each age from 65 through 84 under a $30,000 pension: 20 × $30,000 = $600,000. An option paying $27,000 from ages 62 through 84 makes 23 payments, totaling $621,000. Its three early payments add $81,000, while the $3,000 reduction across 20 later payments saves $60,000. On these simplified assumptions, the early option costs $21,000 more.

Watch the distinction: A 10% reduction in annual benefit is not necessarily an actuarially neutral early-retirement adjustment. Count the additional payments and value both full streams.

Reference: [7]

48. Compare pension cost methods

An actuarial cost method assigns the present value of projected benefits across periods of service. The portion attributed to service already rendered forms an actuarial accrued liability; the portion assigned to the current period contributes to normal cost. Those terms concern timing of attribution, not a change in the total benefits ultimately payable. Different methods may project pay to retirement, value benefits earned to date, or spread projected cost using a chosen pattern. They therefore produce different liability and cost paths even for the same participants and assumptions. Examine the valuation purpose before selecting a method: funding, accounting, and plan termination can ask different questions. Keep the method’s attribution rule separate from asset valuation and from later experience gains or losses. A stable normal cost can coexist with a volatile funded position when assets or assumptions change.

Worked example: Suppose a member’s projected benefit has a present value of 120 valuation units today and is attributed evenly across 20 service years. After 10 completed years, 10/20 × 120 = 60 units are assigned to past service. The next year’s attribution at this valuation date is 120/20 = 6 units. This simplified split holds the present value fixed; an actual later valuation would update it for interest, experience, and assumptions.

Watch the distinction: Normal cost is not the year’s cash benefit payment, and accrued liability is not the whole projected benefit value. Both depend on the attribution method.

Reference: [7]

49. Set valuation assumptions

Assumptions translate a promised benefit into expected timing and amount. Mortality governs survival to each payment; retirement and termination assumptions govern when benefits begin or stop accruing; salary growth matters for pay-related formulas; discount rates convert future payments into present value. Select them for the valuation’s purpose and the plan’s population, then assess whether they are internally consistent. For example, a later retirement assumption can increase service accruals yet postpone payments, so its net effect requires projection. Distinguish a best-estimate assumption from a deliberate margin where the applicable basis calls for one. Compare actual experience with expected counts and amounts, but avoid treating a small, volatile sample as decisive. Sensitivities isolate drivers by changing one assumption while holding others fixed. A final recommendation should explain both the observed evidence and the liability effect of plausible alternatives.

Worked example: Suppose a certain payment of 100 units falls due exactly ten years from now. At a 4% annual effective discount rate, its value is 100/1.04^10 = 67.56 units. At 3%, it is 100/1.03^10 = 74.41 units, an increase of 6.85. This isolates discounting because payment amount and timing are fixed. A pension with survival-dependent payments also needs mortality assumptions.

Watch the distinction: A lower discount rate generally raises the present value of fixed future payments, but it does not replace analysis of mortality, retirement, or salary assumptions.

Reference: [7]

50. Read pension accounting movements

A pension obligation roll-forward explains why the closing measurement differs from the opening one. Current service cost adds benefits attributed to work during the period. Interest cost unwinds one period of discounting on the opening obligation. Benefits paid reduce the obligation because promised cash flows have been settled. Remeasurement captures changes arising from experience or revised assumptions, subject to the accounting framework used. Plan amendments, settlements, currency effects, and other specified events may need separate lines. Reconcile the arithmetic first, then classify each movement according to IAS 19 or U.S. GAAP; the location of a movement in profit, loss, or other comprehensive income can differ by framework. An unexplained balancing figure is a diagnostic cue, not permission to label the difference service cost. Check opening and closing dates, benefit payments, and whether figures are gross or net of plan assets.

Worked example: A hypothetical obligation opens at 100 million currency units. Service cost is 5 million, interest cost 4 million, and benefits paid 3 million. Before remeasurement or other changes, closing obligation is 100 + 5 + 4 − 3 = 106 million. If the reported amount is 112 million, a further 6 million increase needs explanation. If revised mortality is the only additional item, classify the 6 million as remeasurement, not service cost.

Watch the distinction: The difference between opening and closing obligation is not automatically pension expense. Benefits paid, interest, service, and remeasurement have distinct causes and reporting treatment.

Reference: [7]

51. Match pension assets to liabilities

Funded status is asset value minus liability value, so rate risk depends on both sides. Modified duration approximates each side’s percentage price change for a small parallel yield move: change in value is approximately negative duration times yield change times current value. To compare exposures, multiply each duration by its value; matching duration numbers alone is insufficient when assets and liabilities differ in size. A longer liability duration than asset duration usually means falling yields worsen funded status. Liability-responsive investing can add long bonds or interest-rate derivatives to raise asset sensitivity. Yet parallel duration matching does not eliminate curve-shape, inflation, credit, longevity, or liquidity risk. Test how benefit payments and collateral calls would be met after a rate shock. Recalculate sensitivities as yields, membership, and asset mix change, since duration is a local approximation.

Worked example: Assume assets of 90 million have modified duration 4 and liabilities of 100 million have duration 10. For a small parallel 1% yield fall, asset value rises approximately 90 × 4 × 0.01 = 3.6 million; liability value rises approximately 100 × 10 × 0.01 = 10 million. The initial 10 million deficit worsens by 10 − 3.6 = 6.4 million, to about 16.4 million. Greater asset rate sensitivity could reduce that exposure.

Watch the distinction: Owning long bonds does not establish a duration match. Compare value-weighted asset and liability sensitivities, then check liquidity and nonparallel yield changes.

Reference: [7]

52. Separate Canadian pension valuation bases

A Canadian going-concern valuation asks how an ongoing plan may meet benefits through its future operation. A solvency or hypothetical wind-up valuation asks a different question: what the promised benefits would cost on a specified discontinuance basis. Assumptions, settlement methods, and asset treatment can differ, so the same plan can show surplus on one basis and shortfall on another. A full wind-up requires its own facts and valuation date. Commuted values and asset transfers also have distinct purposes and applicable bases; neither can safely be inferred from a plan-wide going-concern surplus. Start by naming the decision, such as ongoing contributions, benefit security on discontinuance, or a member transfer. Then choose the matching liability and asset measures and calculate the difference. For Canadian work, identify the governing jurisdiction and valuation date before applying detailed rules, because they can vary.

Worked example: A hypothetical Canadian plan holds 95 million in assets. Its going-concern liability is 90 million, giving 95 − 90 = 5 million surplus. Its hypothetical wind-up liability is 105 million, giving 95 − 105 = −10 million, a 10 million shortfall. The positive ongoing result does not show that discontinuance benefits are fully secured. A proposed member asset transfer needs its own valuation basis and transfer analysis.

Watch the distinction: Do not use a going-concern surplus as a substitute for a solvency result or an individual commuted value. Each answers a different question.

Reference: [7]

Cross Practice

53. Test a growth strategy

Growth can come from building capabilities internally or buying them. Examine the competitive setting first: rivalry, buyers’ and suppliers’ bargaining power, substitutes, and entry barriers affect how much new revenue can be retained. Value-chain analysis then identifies where the insurer actually creates margin, such as distribution, underwriting, administration, or claims management. An acquisition may buy speed or access but also brings purchase price, integration costs, and execution risk. Organic growth may take longer while allowing tighter control. Build a budget that separates incremental revenue, variable costs, fixed costs, capital needs, and timing. Compare alternatives on the same horizon and discount basis, including synergies only where an operating change can deliver them. Set milestones that would reveal a failed assumption early, such as conversion rate or acquisition cost, and revisit the strategy when they diverge.

Worked example: Suppose a channel adds 20 million of first-year revenue at a 30% contribution margin, or 6 million before fixed costs. Building costs 4 million, leaving 2 million. Buying a distributor costs 9 million but saves 3 million in operating costs, leaving 6 − 9 + 3 = 0 million. Organic growth wins on first-year contribution. An acquisition case needs credible later cash flows or strategic benefits to justify its upfront disadvantage.

Watch the distinction: Higher acquired revenue is not proof of greater value. Compare contribution after incremental costs, integration expense, timing, and the competitive durability of any synergy.

Reference: [8]

54. Optimize under real constraints

A constrained allocation problem needs decision variables, an objective, and explicit limits. If x and y represent staff-days assigned to two projects, maximize their total expected contribution subject to available days and each project’s capacity. For a linear objective, use the highest marginal contribution first while its limit and other constraints allow; with interacting risks or diminishing returns, that ranking may change. State whether days are divisible and whether benefits are certain, because integer staffing and uncertain outcomes may call for different models. The mathematical answer is only as useful as the objective supplied. A manager measured on project volume might overstate its contribution or omit customer harm, capital strain, or future workload. Challenge the inputs, run sensitivities, and compare the recommended allocation with operational feasibility before making a decision.

Worked example: A firm has 10 staff-days. Project A earns 8 value units per day for at most 6 days; B earns 6 per day for at most 8. Assign 6 days to A and 4 to B, earning 6 × 8 + 4 × 6 = 72 units. Moving one day from A to B reduces value by 2 units. If a manager is rewarded solely for B’s volume, their preferred allocation may not maximize firm value.

Watch the distinction: An optimized answer is not objective if its inputs omit costs or reflect incentives. Check the stated constraints and value measure before accepting the allocation.

Reference: [8]

55. Design nested insurance models

Nested insurance models use outer scenarios to describe possible future economic states and inner simulations to value liabilities within each state. For a market-sensitive guarantee, the inner calculation may require many paths because future option payoffs depend on uncertain markets and policyholder behavior. This creates a large computational burden when every outer state receives a full inner valuation. A proxy model learns the relationship between selected outer-state features and full-model values, then predicts values for the remaining states. It is useful only where it reproduces the decision-relevant features: tail values, nonlinear guarantee responses, dynamic lapses, and changes near exercise boundaries. Select training states across both ordinary and stressed conditions. Validate on withheld states using errors in capital or hedge metrics, not only average dollar error. Document where extrapolation is unreliable and rerun the full model when conditions move beyond the training range.

Worked example: Suppose 1,000 outer market scenarios each require 500 inner valuation paths. Full calculation uses 1,000 × 500 = 500,000 inner paths. Training a proxy on 100 outer states uses 100 × 500 = 50,000, then evaluates the other 900 states cheaply. That is a 450,000-path reduction in full inner work. Before using its capital result, compare proxy and full-model values on withheld stressed states, especially where lapses change sharply.

Watch the distinction: A proxy with small average error can still miss the tail that drives capital. Test stressed states and behavioral changes before relying on its speed.

Reference: [8]

56. Govern scenario and predictive models

Model governance begins with a defined use: pricing, reserving, capital, or forecasting may require different accuracy and scenario behavior. Development records data sources, assumptions, calibration targets, implementation, and known limits. Validation then challenges more than the fitted inputs. For an economic scenario generator, test the starting yield curve, distributions over future horizons, correlations, and liability results. For a predictive model, examine out-of-sample performance, stability across populations, and whether variables create unintended bias. Independent review should reproduce key outputs and probe sensitive assumptions. Change control records who altered the model, why, what testing followed, and which decisions use the new version. Monitoring compares realized experience and market conditions with the model’s expected range. If a material weakness appears, quantify its effect and restrict the model’s use until corrected.

Worked example: A hypothetical rate generator exactly matches today’s yield curve but produces no negative rates. A guarantee becomes costly when rates fall below zero, so curve fit alone misses a material exposure. Run a sensitivity with negative-rate paths and compare guarantee values. If the difference affects the capital decision, revise the generator or limit its approved use, record the weakness, and retest after the change.

Watch the distinction: Matching today’s observed prices does not validate all future scenarios. A calibrated model can still omit outcomes that drive the liability being measured.

Reference: [8]

57. Value long-duration health claims

Long-duration health liabilities depend on whether a claim has occurred. A policy reserve addresses future claims arising under coverage still in force, along with future premiums and expenses under the chosen valuation basis. A claim reserve addresses obligations from claims already incurred. For an open disability or long-term-care claim, projected payments not yet due are valued using continuation, recovery, mortality, benefit limits, offsets, and discounting assumptions. Amounts already due but unsettled belong in the in-course-of-settlement component; incurred claims not yet reported require a separate estimate. This classification prevents counting the same future payment as both a policy claim and an open-claim liability. Diagnose reserve movements by separating new incidence, claim termination, benefit growth, and payment timing. The distinction also helps when comparing with general insurance reserving, where reported and unreported claim development likewise require separate attention.

Worked example: Assume an existing disability claimant is expected to receive $2,000 at each of the next 12 month ends, with certain continuation and a hypothetical 0% monthly discount rate. The present value of amounts not yet due is 12 × $2,000 = $24,000. It belongs to the incurred claim. Add any payments already due but unsettled and a separate incurred-but-unreported estimate; future incidence among claim-free policyholders belongs in policy-reserve projections.

Watch the distinction: A payment scheduled next year is not necessarily a future claim. If the disability began already, the unpaid amount is part of the existing claim liability.

Reference: [8]

58. Examine advanced life reinsurance

Life reinsurance transfers specified insurance economics, but treaty structure determines where assets sit and how cash settles. In ordinary coinsurance, a proportional share of premiums, claims, and supporting reserves generally passes to the reinsurer under the contract. In modified coinsurance or funds-withheld arrangements, the cedant may retain supporting assets while owing investment returns or settlement amounts to the reinsurer. That changes custody, collateral, credit, and liquidity questions even if mortality or longevity risk is ceded. Read the treaty’s risk-sharing, funds-withheld balance, recapture, insolvency, and collateral provisions together. A projected capital benefit must be compared with counterparty exposure and the cash needed if the treaty ends. Under recapture, the cedant may have to reestablish reserves or settle a balance quickly. Jurisdiction-specific accounting and credit-for-reinsurance treatment affect reported relief, so economic transfer and reported capital relief should be assessed separately.

Worked example: Assume a cedant transfers insurance risk associated with 80 million of reserves but retains 80 million of supporting assets under a hypothetical funds-withheld treaty. On recapture, a specified settlement of 85 million becomes due. If only the retained 80 million is readily available, the immediate funding gap is 85 − 80 = 5 million. The risk transfer does not remove that liquidity exposure; assess collateral and settlement terms before counting the full benefit.

Watch the distinction: Ceding reserve risk does not mean the cedant has sent the supporting assets away or eliminated counterparty risk. Trace custody, settlement, and recapture cash flows.

Reference: [8]

59. Hedge key-rate mismatches

Overall duration summarizes sensitivity to a parallel change in yields. Key-rate sensitivity instead measures the effect of moving one selected point on the yield curve while holding the others according to a defined method. This reveals mismatches hidden by equal total durations. Express asset, liability, and hedge effects with one sign convention, such as change in surplus per basis-point rise. Sum them at each key rate before sizing a hedge. Divide the opposite of the net exposure by the hedge’s per-unit sensitivity, then check other maturities: an instrument that fixes the 20-year point may add unwanted 10-year exposure. The calculation is local and assumes small moves; convexity, basis between discount and hedge curves, trading costs, collateral, and counterparty risk remain. For long liabilities, test curve steepening and flattening alongside the single-point shock.

Worked example: At the 20-year key rate, suppose a one-basis-point rise reduces surplus by $40,000 through liabilities and increases it by $10,000 through assets. Net sensitivity is −$30,000 per basis point. Each hedge unit adds $15,000 for the same rise, so two units add $30,000 and bring the local net to zero. Check their effects at other key rates and their collateral needs before executing.

Watch the distinction: Matching total duration can leave a large curve-shape mismatch. Use signed key-rate exposures, and check what the proposed hedge does outside its target maturity.

Reference: [8]

60. Stress asset-liability strategy

An asset-liability strategy must survive cash demands as well as changes in reported value. A rate shock can move assets and liabilities favorably in economic terms yet require immediate derivative collateral. Surrenders or benefit payments create another cash need, sometimes when selling assets is costly. Build a dated liquidity ladder: opening cash and assets reliably convertible to cash, plus expected inflows, minus contractual outflows, behavior-driven withdrawals, and collateral. Stress interacting factors together because rate moves may trigger both collateral calls and lapses. Distinguish assets available in the right legal entity and currency from group-wide holdings that cannot be moved promptly. Compare the lowest projected balance with an internal operating buffer, then identify actions such as holding liquid securities, arranging committed funding, reducing hedge collateral volatility, or slowing illiquid purchases. A positive ending balance alone can hide an earlier intraday or interim shortfall.

Worked example: A hypothetical insurer starts with 30 million of usable liquid assets. A rate shock requires 18 million of collateral, surrenders require 15 million, and scheduled inflows supply 5 million during the same period. Closing liquidity is 30 − 18 − 15 + 5 = 2 million. The arithmetic is positive, but the narrow buffer and payment timing warrant a liquidity action before adding illiquid assets.

Watch the distinction: An ALM strategy is not workable merely because projected surplus stays positive. Collateral and withdrawals may require cash before assets can be sold or inflows arrive.

Reference: [8]

Official references and further reading

All study guides

Choose your exam version

FSA CFE 101

FSA CFE 201

FSA GH 101

FSA GH 201-U

FSA GH 201-C

FSA GH 301

FSA GI 101

FSA GI 201

FSA GI 301

FSA GI 302

FSA ILA 101

FSA ILA 201-U

FSA ILA 201-I

FSA INV 101

FSA INV 201

FSA RET 101

FSA RET 201

FSA RET 301

FSA CP 311

FSA CP 312

FSA CP 321

FSA CP 341

FSA CP 351

Corporate Finance and ERM

1. Assign risk ownership and oversight

2. Turn emerging risks into indicators

3. Read tail risk measures

4. Aggregate dependent risks

5. Trace a stress through the balance sheet

6. Choose a risk response

7. Allocate capital to decisions

8. Value a corporate project

9. Challenge reported performance

Group and Health Insurance

10. Match health benefits to risk

11. Build a medical manual rate

12. Blend group experience and manual cost

13. Select an employer funding approach

14. Estimate unpaid health claims

15. Interpret U.S. health insurer capital

16. Read Canadian insurance results

17. Evaluate provider payment and care programs

18. Normalize health risk before comparison

General Insurance

19. Prepare general insurance exposure

20. Select ultimate claim methods

21. Correct distorted claim development

22. Separate frequency and severity trend

23. Price deductibles and limits

24. Price a reinsurance layer

25. Measure reserve uncertainty

26. Connect legal change to claim cost

27. Read a U.S. insurer's reserve opinion

Individual Life and Annuities

28. Map life and annuity guarantees

29. Test life product profitability

30. Set behavior assumptions from experience

31. Manage non-guaranteed benefits

32. Build a life reserve projection

33. Separate U.S. GAAP liability components

34. Interpret IFRS 17 service margin

35. Distinguish capital measures

36. Evaluate life reinsurance cash flows

Investment

37. Inspect fixed-income cash-flow risk

38. Value illiquid alternatives

39. Turn liabilities into an investment policy

40. Measure bond rate exposure

41. Attribute portfolio performance

42. Estimate expected credit loss

43. Price by replication

44. Test a dynamic hedge

Retirement Benefits

45. Choose a retirement plan design

46. Measure retirement income adequacy

47. Price embedded retirement options

48. Compare pension cost methods

49. Set valuation assumptions

50. Read pension accounting movements