Use this guide to connect ratemaking and unpaid claims formulas with the insurance data and assumptions behind them. Each concept explains a specific distinction or calculation, resolves an original example and identifies an error to avoid. Work through the foundations first, then compare how different methods respond to the same changes in exposure, claim experience and development.
Insurance data and measurement foundations
1. Choosing an exposure unit
An exposure unit measures the amount of insured risk underlying a rate. Choose a denominator that corresponds to the coverage and the period during which risk exists. Claim frequency and loss cost become misleading when their denominators change definition between historical experience and the prospective rating calculation.
Worked example: A portfolio has 240 claims across 12,000 vehicle-years. Its frequency is 240 / 12,000 = 0.02 claims per vehicle-year, or two claims per 100 vehicle-years.
Mistake to avoid: Dividing annual claims by a year-end vehicle count without checking whether it represents annual exposure.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
2. Written and earned exposure
Written exposure records coverage issued; earned exposure records the portion of coverage elapsed. For uniformly exposed annual policies, earning follows elapsed coverage time. Match losses to earned exposure when estimating historical loss costs, and account for cancellations or other changes that alter the actual time at risk.
Worked example: Eighty annual policies begin July 1. Ignoring cancellations, six months of coverage through December 31 produce 80 × 6 / 12 = 40 policy-years of earned exposure.
Mistake to avoid: Treating all 80 written policies as 80 earned policy-years during that calendar year.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
3. Written and earned premium
Written premium reflects premium booked for issued coverage; earned premium reflects the portion associated with coverage already provided. Under uniform earning, an annual premium earns proportionally over the policy term. Premium earning and exposure earning must use compatible coverage periods before calculating average premium or a loss ratio.
Worked example: An annual policy has premium of $9,000 and three months of coverage have elapsed. With uniform earning, earned premium is $2,250 and the remaining $6,750 is unearned.
Mistake to avoid: Using the full written premium against only three months of earned losses.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
4. Accident, policy and calendar periods
Accident periods assign claims by occurrence date. Policy periods assign experience by policy inception date. Calendar periods assign transactions or accounting movements by when they occur. These views answer different questions, so identify the organizing date before combining premium, exposure and claims or interpreting a development pattern.
Worked example: A policy starts November 2024, a loss occurs February 2025 and payment occurs August 2026. The claim belongs to policy year 2024, accident year 2025 and payment calendar year 2026.
Mistake to avoid: Assuming a claim's policy year and accident year must match.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
5. Calendar-year incurred loss movements
For a consistent portfolio, calendar-year incurred losses equal payments during the year plus closing case reserves minus opening case reserves. This measures accounting activity across multiple accident years. It can reflect revisions to older claims, making it different from the estimated ultimate cost of accidents occurring during that year.
Worked example: Payments are $700,000, opening case reserves are $900,000 and closing case reserves are $1,050,000. Calendar-year incurred losses are $700,000 + $1,050,000 − $900,000 = $850,000.
Mistake to avoid: Interpreting all calendar-year incurred losses as costs of that year's accidents.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
6. Aggregation and changing business mix
Aggregate experience combines both within-segment performance and the distribution of business across segments. A change in average loss cost can arise from a changing mix even when every segment's loss cost remains constant. Compare segment experience and exposure weights before attributing an aggregate movement to inflation or deteriorating risk.
Worked example: Loss costs remain $100 and $300 in two classes. Moving the high-cost class from 20% to 40% of exposure raises the average from $140 to $180 without any within-class increase.
Mistake to avoid: Applying the entire aggregate increase as a trend assumption to every class.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
7. Paid claims, case reserves and reported incurred
Paid claims measure amounts already disbursed. Case reserves estimate remaining amounts on reported claims. Under the stated convention, reported incurred claims equal paid claims plus case reserves. Reported incurred is an estimate at a valuation date, so changes can arise from payments, new reports or revised case estimates.
Worked example: A claim has $18,000 paid and a $12,000 case reserve. Reported incurred is $30,000. A further $4,000 payment with an equal reserve reduction leaves reported incurred unchanged.
Mistake to avoid: Adding payments to reported incurred without recognizing the corresponding case-reserve movement.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
8. Reconciling data on a consistent basis
Before calculating indications, reconcile extracts to an independent control using the same valuation date, coverage, currency and gross or net basis. Investigate excluded transactions, duplicates and recovery handling. A reconciliation difference is evidence about data preparation; selecting a different actuarial assumption does not resolve an unexplained extraction error.
Worked example: A claims extract totals $4.80 million against a $5.00 million control. Identifying $0.20 million of omitted reopened-claim transactions reconciles the amounts before development factors are calculated.
Mistake to avoid: Increasing development factors to compensate for missing transactions instead of correcting the extract.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
Premium and loss adjustments
9. Compounding rate-level changes
Rate changes accumulate multiplicatively. A rate-level index tracks the relative premium produced by successive changes while holding the underlying risk constant. Calculate the current index and historical indices on one base before adjusting experience premium. Adding percentage changes generally gives the wrong cumulative rate level.
Worked example: Starting from index 1.00, a 10% increase followed by a 5% decrease produces 1.00 × 1.10 × 0.95 = 1.045, a net increase of 4.5%.
Mistake to avoid: Reporting a 5% net increase by simply adding 10% and −5%.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
10. On-level premium from historical rate mixtures
On-leveling estimates the premium historical risks would produce at a selected rate level. When the experience contains multiple rate levels, calculate the appropriate exposure or earning weights first. The adjustment factor is the target index divided by the weighted historical index, assuming the index captures the relevant rate changes.
Worked example: Half the earned exposure is at index 1.00 and half at 1.20. The average index is 1.10. Moving $550,000 of earned premium to index 1.20 gives $600,000.
Mistake to avoid: Averaging separate on-level factors when the required calculation uses a weighted rate-level index.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
11. Parallelogram earning assumptions
A parallelogram calculation estimates the portion of earned premium associated with each rate level using policy issuance and earning assumptions. Under uniformly written annual policies with uniform earning, a rate change affects earned premium gradually. Uneven writings, different policy terms or unusual cancellations can require a more detailed calculation.
Worked example: A new-business rate change takes effect July 1. Under the stated annual-policy assumptions, the new level represents 12.5% of that calendar year's earned exposure, rather than 50%.
Mistake to avoid: Assigning half the year's earned premium to the new rate solely because the change occurred midyear.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
12. Loss trend and the projection interval
Loss trend projects historical costs to the cost level expected during the prospective coverage period. Measure the interval between appropriate average dates, rather than automatically using whole calendar years. Keep the trend assumption distinct from claim development, which addresses emergence of costs for existing accident periods.
Worked example: Historical losses average July 1, 2024, and prospective losses average July 1, 2027. At 4% annual trend, $500,000 projects to $500,000 × 1.04³ = $562,432.
Mistake to avoid: Using the policy effective date when the calculation calls for the average future accident date.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
13. Premium trend versus rate-level adjustment
Premium trend captures changes in average premium associated with exposure characteristics or a changing rating base, apart from explicit rate revisions. On-leveling addresses rate revisions. Separate the two so the same increase is not counted twice, and ensure premium trend represents the risks expected in the prospective period.
Worked example: At unchanged rates, average premium rises from $800 to $840 because insured values increase. This is a 5% premium trend effect; a separate 6% rate revision belongs in on-leveling.
Mistake to avoid: Including the 6% rate revision in both premium trend and the rate-level adjustment.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
14. Developing losses before ratemaking
Ratemaking generally needs an estimate of ultimate historical claims, rather than only amounts reported or paid so far. Apply development assumptions appropriate to the data basis and maturity, then project the resulting cost level to the prospective period. Development and trend serve separate purposes even when both increase the estimate.
Worked example: Reported losses are $400,000 with a selected ultimate development factor of 1.25. Ultimate losses are $500,000; a further 10% prospective trend adjustment gives $550,000.
Mistake to avoid: Treating the development factor as an inflation assumption or applying it to already ultimate losses.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
15. Large-loss limits and excess provisions
Limiting individual historical losses can reduce volatility in the experience used for pricing. When the prospective coverage still includes amounts above that limit, add a separately estimated excess provision. The limited-loss calculation and excess provision must cover complementary portions of the same insured losses without gaps or overlap.
Worked example: Losses of $20,000, $70,000 and $250,000 total $340,000. Limiting each at $100,000 gives $190,000. A selected $120,000 expected excess provision produces an adjusted estimate of $310,000.
Mistake to avoid: Using limited historical losses as the entire expected cost of unlimited coverage.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
16. Normalizing catastrophe costs
A short experience period may contain unusually high or low catastrophe losses. One approach separates actual catastrophe losses from ordinary experience and substitutes a defensible prospective catastrophe provision. Preserve the definition of catastrophe exposure and coverage; removing an event does not eliminate its expected future cost.
Worked example: Historical losses total $2.4 million, including $0.9 million from a catastrophe. Replacing that amount with a $0.4 million prospective provision yields $1.9 million before other adjustments.
Mistake to avoid: Adding a catastrophe provision while leaving the replaced catastrophe losses in the base.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
17. Claim adjustment expenses in pricing
Claim adjustment expenses are costs of investigating, defending and settling claims. Some are attributable to individual claims; others are shared claim-handling costs. Determine what the loss data already include and estimate omitted components on a compatible basis. Expense terminology alone does not establish how a particular extract is constructed.
Worked example: A projected claim cost of $800,000 includes $60,000 of claim-specific expenses. Adding $40,000 of separately estimated shared handling expense gives $840,000, not $900,000.
Mistake to avoid: Adding claim-specific expenses again because an expense summary lists them separately.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
18. Consistent gross and net recovery bases
Recoveries such as salvage, subrogation and reinsurance can change the loss amount used in pricing. State which recoveries are reflected and pair losses with the corresponding premium or cost basis. Reinsurance recoveries require consideration of reinsurance cost; a lower net claim estimate is not automatically a lower total insurance cost.
Worked example: Gross losses are $1.0 million and expected reinsurance recoveries are $0.2 million. Net losses are $0.8 million; including a $0.15 million reinsurance premium gives $0.95 million before other costs.
Mistake to avoid: Subtracting reinsurance recoveries while omitting the cost of obtaining that protection.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
Rate indications, expenses and profitability
19. Pure premium as expected loss cost
Pure premium is expected claim cost per exposure unit, with any included claim expenses explicitly identified. Divide projected ultimate costs by comparable earned exposure. This gives the loss-cost component of a rate; underwriting expenses and profit provisions must be incorporated separately to obtain a required premium.
Worked example: Projected ultimate claims and included claim expenses are $720,000 across 6,000 policy-years. Pure premium is $120 per policy-year. It is not yet the final charged rate.
Mistake to avoid: Calling the pure premium a complete rate when acquisition expenses and profit are excluded.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
20. Loss-ratio rate indications
The loss-ratio approach compares a projected loss ratio at the current rate level with a permissible loss ratio consistent with expense and profit assumptions. The indicated premium multiplier is projected divided by permissible loss ratio. Both ratios must use compatible definitions, especially for claim expenses and fixed underwriting expenses.
Worked example: A projected loss ratio of 72% and permissible loss ratio of 60% imply a multiplier of 0.72 / 0.60 = 1.20, or a 20% indicated increase.
Mistake to avoid: Reporting the 12-percentage-point ratio difference as the indicated percentage rate change.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
21. Fixed and variable underwriting expenses
Fixed expenses are modeled as amounts that do not vary directly with premium, often expressed per policy or exposure. Variable expenses are modeled as a percentage of premium. Their different behavior matters when premiums change: applying one historical expense ratio to all expenses can misstate the prospective cost.
Worked example: With $500 premium, $40 fixed expense and a 15% variable expense provision, total expense is $115. At $600 premium, unchanged assumptions give $40 + $90 = $130.
Mistake to avoid: Assuming fixed dollar expenses increase automatically by the same percentage as premium.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
22. Projecting expense levels
Project expense dollars and expense percentages according to their underlying drivers. Salary or service-cost inflation may affect fixed expenses, while contractual commissions may remain a percentage of premium. Historical expense ratios can also move because premium changes, so distinguish changes in the numerator from changes in the denominator.
Worked example: Fixed expense of $30 per policy increases 4% annually for two years, reaching $32.448, approximately $32.45. A separately assumed 12% commission remains 12% of prospective premium.
Mistake to avoid: Applying fixed-expense inflation to a commission percentage without evidence that the percentage changes.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
23. Solving for premium with a profit provision
When variable expenses and an underwriting profit provision are percentages of premium, premium appears on both sides of the cost equation. Required premium equals expected losses plus fixed expenses, divided by one minus the variable-expense ratio minus the profit ratio. Use this denominator only for the explicitly stated pricing structure.
Worked example: Expected losses are $140 per exposure, fixed expenses $20, variable expenses 15% and profit 5%. Required premium is $160 / 0.80 = $200.
Mistake to avoid: Adding 20% to $160, which gives $192 and does not fund provisions measured against premium.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
24. Underwriting profit and investment income
Underwriting profit compares premium with claims and underwriting expenses. Investment income is a separate source of return influenced by cash-flow timing and invested assets. If investment income is considered in a pricing analysis, use an explicit financial basis rather than assuming a universal offset to the underwriting provision.
Worked example: Premium of $1,000 less claims of $650 and underwriting expenses of $280 leaves $70 underwriting profit. Separately identified investment income of $25 raises the combined pre-tax contribution to $95.
Mistake to avoid: Presenting a $95 contribution as underwriting profit or assuming investment income is always available at that amount.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
25. Combined ratios and denominator consistency
A combined ratio adds loss and expense ratios under specified definitions. If one component uses earned premium and another uses written premium, their sum does not directly equal total costs divided by one common premium amount. Identify each denominator before translating a ratio into underwriting dollars or comparing portfolios.
Worked example: Losses of $600 and expenses of $300 against $1,000 earned premium give a consistent combined ratio of 90% and underwriting profit of $100.
Mistake to avoid: Inferring exact underwriting dollars from a combined ratio whose components use different premium denominators.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
26. Aggregating segment rate indications
A portfolio rate indication combines segment changes using the premium base affected by those changes. Average indicated percentage changes with current premium weights, or calculate total required premium directly. Exposure weights answer a different question when segment average premiums differ, and can produce the wrong overall premium change.
Worked example: Segment A has $100,000 premium and needs 20% more; segment B has $300,000 and needs 4% less. Required premium is $120,000 + $288,000 = $408,000, a 2% increase.
Mistake to avoid: Taking the unweighted average of 20% and −4% and reporting an 8% portfolio indication.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
Classification and individual risk rating
27. Class relativities and a base rate
A class relativity expresses a class's indicated cost or rate relative to a selected base class. Apply compatible exposure definitions and adjust experience to comparable cost levels before estimating relativities. The base class sets the scale; changing the base should not change the underlying relative price relationships.
Worked example: Comparable loss costs are $120 for the base class and $180 for another class. The second class has relativity 1.50. With a $200 base rate, its rate is $300.
Mistake to avoid: Treating the $60 loss-cost difference as a premium surcharge without considering the rating structure.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
28. Normalizing relativities to preserve premium
New relativities can change aggregate premium even when the base rate remains unchanged. To preserve a selected overall premium level, adjust the base rate for the change in the weighted average rating factor. Use weights representing the portfolio to which the rates will apply, including other relevant rating factors.
Worked example: Equal exposures previously have factors 1.0 and 1.2, averaging 1.1. New factors 1.0 and 1.4 average 1.2. A $100 base becomes $100 × 1.1 / 1.2 = $91.67.
Mistake to avoid: Keeping the base rate unchanged and describing the new relativities as premium-neutral.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
29. Confounding in one-way class comparisons
A one-way comparison can combine a class effect with differences in other risk characteristics. Examine whether the compared classes have similar distributions of coverage, territory or other rating variables. Standardization or an appropriate multivariable analysis can help separate effects; an observed difference alone does not identify its cause.
Worked example: Class A averages $200 in losses and class B $300, but B contains more high-cost territories. After using common territory weights, costs are $240 and $250, suggesting a much smaller class difference.
Mistake to avoid: Assigning the entire unadjusted 1.50 ratio to class membership.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
30. Credibility-weighted estimates
Credibility blends observed experience with a relevant complement: estimate = Z × experience + (1 − Z) × complement. The credibility weight reflects the chosen method and available information. The complement should represent comparable risks and cost levels; blending incompatible estimates does not become sound merely because the weights sum to one.
Worked example: A class loss cost is $150, the relevant complement is $110 and assigned credibility is 0.40. The blended estimate is 0.40 × $150 + 0.60 × $110 = $126.
Mistake to avoid: Using an unrelated portfolio average as the complement without adjusting for systematic risk differences.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
31. Experience rating against expected losses
Experience rating modifies an individual risk's prospective price using its past losses relative to losses expected for its characteristics and exposure. This separates individual experience from class-level pricing. Actual plans can contain additional components; a simple credibility-weighted actual-to-expected ratio illustrates the principle without defining any particular filed plan.
Worked example: Actual losses are $120,000 versus $100,000 expected. With credibility 0.25, an illustrative modifier is 0.25 × 1.20 + 0.75 × 1.00 = 1.05.
Mistake to avoid: Comparing raw losses between risks with substantially different exposure sizes.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
32. Loss limitation in individual risk experience
A large individual claim can make an experience measure volatile. A rating design may limit or separately weight large losses to emphasize more repeatable information. Apply the same loss treatment to actual experience and its expected benchmark; otherwise the resulting modifier can reflect a mismatch rather than the risk's relative performance.
Worked example: Two claims of $15,000 and $180,000 total $195,000. Under an illustrative $50,000 per-claim cap, actual limited losses are $65,000 and must be compared with expected limited losses.
Mistake to avoid: Comparing capped actual losses with an unlimited expected-loss benchmark.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
33. Schedule rating and duplicate adjustments
Schedule rating assesses specified risk characteristics that may not be fully represented by class rates or historical experience. A defensible adjustment identifies the characteristic and its expected cost effect. Consider existing classification and experience adjustments before adding another factor, so one favorable or adverse feature is not counted repeatedly.
Worked example: A base premium is $10,000. An illustrative 5% schedule credit gives $9,500. If the same characteristic already drives a classification discount, review whether another credit duplicates its effect.
Mistake to avoid: Applying an additional credit solely for a feature already fully reflected in the underlying rate.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
34. Interactions in multiplicative rating structures
A multiplicative rating structure assumes each factor's relative effect is stable across combinations unless interaction terms are included. Test that assumption using comparable combinations. An interaction means the combined effect differs from the product of separate effects; it should represent an identifiable pattern rather than compensate for unexplained data problems.
Worked example: Factors of 1.20 and 1.30 predict a combined relativity of 1.56. If the supported combined relativity is 1.80, the additional interaction factor is 1.80 / 1.56 ≈ 1.154.
Mistake to avoid: Adding the separate percentage increases and assuming that matches a multiplicative rate.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
Development methods and reserve diagnostics
35. Development triangle coordinates
A development triangle arranges origin periods by successive ages since origin. Cells on a calendar diagonal correspond to observations associated with the same calendar period under the chosen convention. Future cells are unobserved, not observed zeros. Understanding the coordinates helps distinguish origin-year changes from development-age and calendar-period effects.
Worked example: With annual origin periods and year-end valuations, accident year 2024 at 24 months and accident year 2025 at 12 months both lie on the December 2025 valuation diagonal.
Mistake to avoid: Entering zero in future cells and treating it as evidence that no payments will occur.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
36. Incremental and cumulative claims
Incremental claims measure activity within one development interval. Cumulative claims sum activity through an age. Most standard age-to-age development ratios use cumulative values, while incremental data are useful for examining payment timing. Establish the data form before calculating ratios, and investigate negative increments rather than assuming they are impossible.
Worked example: Incremental payments are $90,000, $60,000 and $25,000. Cumulative payments are $90,000, $150,000 and $175,000. The third increment is $175,000 − $150,000 = $25,000.
Mistake to avoid: Calculating a cumulative development factor directly from consecutive incremental payments.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
37. Volume-weighted age-to-age factors
A volume-weighted development factor divides total cumulative claims at the later age by total claims at the earlier age for matched origin periods. It weights individual ratios by earlier-age claim amounts. Include only periods observed at both ages, and check whether one large period dominates the result.
Worked example: Matched earlier amounts are $100,000 and $300,000; later amounts are $150,000 and $360,000. The factor is $510,000 / $400,000 = 1.275, rather than the simple average 1.35.
Mistake to avoid: Including an unmatched recent origin period in the denominator.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
38. Selecting factors from competing averages
A calculated average is evidence for a selection, not an automatic answer. Compare recent and longer histories, volume-weighted and simple averages, and operational explanations for unusual observations. Removing a factor should have a defensible reason; a high factor may represent a genuine recurring pattern rather than a data error.
Worked example: Historical factors are 1.10, 1.12 and 1.80. If the last reflects a documented extraction error, correcting the source is appropriate. If it reflects valid delayed reporting, excluding it requires another justification.
Mistake to avoid: Removing every unfavorable factor merely to obtain a smoother or lower reserve.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
39. Cumulative factors and tail development
A cumulative development factor multiplies all selected remaining age-to-age factors and any tail factor. The tail represents development beyond the last modeled age. Even a modest tail can materially affect ultimate claims, so state its basis and assess sensitivity rather than silently treating the oldest observed age as fully developed.
Worked example: Remaining factors of 1.20 and 1.05 with a 1.02 tail produce 1.2852. Applied to $500,000, estimated ultimate claims are $642,600.
Mistake to avoid: Adding factor excesses or omitting a tail because the observed triangle ends.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
40. Chain-ladder projection on paid and incurred bases
Chain ladder multiplies current cumulative claims by a selected cumulative development factor. Paid and reported-incurred projections can produce different ultimate estimates because their emergence patterns differ. For either basis, total unpaid claims equal selected ultimate minus paid claims; subtracting reported incurred instead gives a different reserve component.
Worked example: Paid claims of $400,000 with factor 1.50 imply $600,000 ultimate and $200,000 unpaid. Reported incurred of $500,000 with factor 1.20 also implies $600,000 ultimate.
Mistake to avoid: Subtracting reported incurred when the requested quantity is total unpaid claims.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
41. Case reserves and aggregate IBNR
When reported incurred equals paid plus case reserves, unpaid claims can be separated into case reserves and the difference between ultimate and reported incurred. That difference is often called aggregate IBNR and may include development on already reported claims. State the convention because strictly unreported claims are only one possible component.
Worked example: Ultimate claims are $900,000, paid claims $550,000 and case reserves $250,000. Unpaid claims are $350,000; aggregate IBNR is $100,000.
Mistake to avoid: Assuming the entire $100,000 relates exclusively to claims that have not yet been reported.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
42. Converting development factors to emergence percentages
Under a selected development pattern, the estimated proportion developed is the reciprocal of the cumulative development factor. The undeveloped proportion is one minus that reciprocal. Interpret developed as paid or reported according to the data basis; these percentages are method assumptions, not measurements of every claim's individual settlement progress.
Worked example: A paid cumulative development factor of 2.50 implies 40% paid and 60% unpaid. It does not imply that 150% of ultimate remains unpaid.
Mistake to avoid: Using the factor minus one as the proportion of ultimate claims still unpaid.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
43. Diagnosing changes in case-reserve strength
Stronger case estimates can increase reported incurred claims without a corresponding change in underlying ultimate costs. Compare average case reserves, paid development and other operational indicators before applying historical incurred factors. A new reserving practice can make old factors inappropriate because the starting reported amounts now contain more estimated future cost.
Worked example: Paid claims remain $2 million while case reserves increase from $1 million to $1.4 million after a reserving review. Reported incurred rises 13.3%, which does not by itself establish higher ultimate losses.
Mistake to avoid: Applying an unchanged incurred factor without examining the documented case-reserving change.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
44. Diagnosing changes in settlement speed
Faster settlements can raise paid claims at a given age while leaving ultimate claims unchanged. Examine closure rates, payment timing and reported-incurred patterns together. Historical paid development factors may then overproject recent origin periods because more of their eventual cost has already been paid at the valuation date.
Worked example: Historically, 12-month paid claims were 40% of ultimate, implying factor 2.50. A credible operational change raises that proportion to 50%, implying 2.00 for the same ultimate cost.
Mistake to avoid: Interpreting faster payments automatically as worsening claim severity.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
45. Segmentation for development estimation
Segments should have sufficiently similar reporting and settlement behavior for a shared development pattern to be meaningful. Aggregating unlike business can distort factors when the business mix changes. Excessive subdivision creates unstable estimates, so balance operational distinctions against data volume and consider relevant broader information for sparse segments.
Worked example: Two segments are typically 80% and 30% paid at 12 months. Increasing the slower segment's share reduces aggregate maturity even if neither segment's own payment pattern changes.
Mistake to avoid: Applying an old aggregate pattern to a portfolio whose segment mix has materially shifted.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
46. Calendar-period effects in claims triangles
A calendar-period event can affect several origin periods simultaneously. Inflation, claim-handling changes or a common reporting disruption may appear along a triangle diagonal rather than within one row. Examine diagonal patterns before attributing the change solely to accident-year quality or development-age behavior.
Worked example: Payments are unexpectedly high across three different origin years during calendar 2025. A shared increase in repair costs is a plausible calendar explanation that merits investigation across those rows.
Mistake to avoid: Treating every affected row as an independent deterioration and overlooking the common calendar driver.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
47. Backtesting with historical information sets
A reserve backtest recreates an earlier valuation using only information then available and compares predictions with later emergence. This tests forecasting behavior rather than merely historical fit. Compare errors by maturity and segment, and distinguish random variation from persistent bias or a known subsequent change in the claims environment.
Worked example: Using only December 2023 data, a method predicts $200,000 of 2024 payments. Actual payments are $230,000, giving a $30,000 underprediction for that interval.
Mistake to avoid: Using factors selected with 2024 observations when claiming to test a December 2023 forecast.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
48. Reserve rollforwards for a fixed claim cohort
For an unchanged claim cohort, closing unpaid liability equals opening unpaid liability minus intervening payments plus the change in estimated ultimate claims. This separates cash settlement from estimate revisions. New accident periods, portfolio transfers and other scope changes must be excluded or shown separately for the reconciliation to remain interpretable.
Worked example: Opening unpaid liability is $800,000, payments are $300,000 and estimated ultimate rises $70,000. Closing unpaid liability is $570,000.
Mistake to avoid: Describing the $230,000 reserve decline as favorable development without accounting for the $300,000 paid.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
Expected claims, Bornhuetter-Ferguson, Cape Cod and frequency-severity
49. Expected ultimate claims from a loss ratio
An expected-loss-ratio estimate multiplies a selected expected loss ratio by premium on a compatible basis. It supplies an expectation independent of the period's observed development to the extent the assumption is independently selected. Rate changes, coverage differences and cost trends must be reflected before transferring a ratio between periods.
Worked example: Comparable earned premium is $2 million and the selected expected loss ratio is 65%. Expected ultimate claims are $1.3 million, before any separately excluded claim expenses.
Mistake to avoid: Applying a historical ratio to changed premium levels without checking whether its cost relationship remains relevant.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
50. Paid Bornhuetter-Ferguson estimation
Paid Bornhuetter-Ferguson adds observed paid claims to expected ultimate claims multiplied by the assumed unpaid proportion. Thus the unpaid estimate depends on the expected cost and payment pattern, rather than directly scaling current paid claims. This can stabilize immature estimates, but an unsuitable expected-loss assumption can still produce bias.
Worked example: Paid claims are $300,000, expected ultimate $800,000 and the paid proportion 40%. Unpaid claims are $480,000, giving estimated ultimate of $780,000.
Mistake to avoid: Forcing the final ultimate estimate to equal the initial $800,000 expectation.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
51. Reported-incurred Bornhuetter-Ferguson estimation
Reported-incurred Bornhuetter-Ferguson adds reported claims to expected ultimate multiplied by the unreported proportion. The added amount estimates aggregate IBNR under the stated convention. To obtain total unpaid claims, subtract paid claims from the resulting ultimate estimate or add existing case reserves to the estimated aggregate IBNR.
Worked example: Reported claims are $600,000, paid $350,000, expected ultimate $1 million and the reported proportion 70%. Ultimate is $900,000; aggregate IBNR is $300,000 and unpaid claims $550,000.
Mistake to avoid: Calling the $300,000 aggregate IBNR estimate the entire unpaid liability.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
52. How chain ladder and Bornhuetter-Ferguson react
At fixed development assumptions, chain-ladder ultimate moves proportionally with observed claims. Bornhuetter-Ferguson ultimate moves dollar-for-dollar with observed claims when expected ultimate remains fixed. The methods therefore respond differently to unusually high or low emergence; whether that difference is useful depends on whether emergence signals changed ultimate costs.
Worked example: With factor 2.00, paid claims rising from $100,000 to $120,000 raise chain-ladder ultimate by $40,000. Paid Bornhuetter-Ferguson ultimate rises $20,000 if its expected unpaid component is unchanged.
Mistake to avoid: Assuming the more stable estimate is automatically more accurate when actual risk costs have changed.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
53. Cape Cod's development-adjusted premium denominator
In a basic reported-claims Cape Cod calculation, estimate a common expected loss ratio by dividing total reported claims by premium weighted by reported proportions. Use that ratio in a Bornhuetter-Ferguson structure. The denominator represents the premium associated with emerged claims, not the full premium for all included origin periods.
Worked example: Premiums are $1,000 and $2,000, reported proportions 80% and 40%, and reported claims total $1,040. The ratio is $1,040 / ($800 + $800) = 65%.
Mistake to avoid: Dividing reported claims by full premium and treating the resulting immature ratio as ultimate.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
54. Making Cape Cod periods comparable
Pooling periods under one expected loss ratio requires a comparable premium and loss basis. Rate changes, loss trend and business mix can otherwise create different underlying ratios. Adjust inputs consistently to the selected reference basis, or segment the calculation when a common ratio is not a defensible representation.
Worked example: Two equal-risk cohorts each have $1,000 historical premium, but the older cohort's rates are half the selected reference level. Its on-level premium becomes $2,000 before constructing a comparable denominator.
Mistake to avoid: Pooling raw premium across materially different rate levels and interpreting the fitted ratio as a common cost relationship.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
55. Projecting ultimate reported claim counts
Claim counts can be developed separately from monetary claims to estimate ultimate claim frequency or unreported counts. Define a count consistently, including treatment of reopened, closed-without-payment and duplicate records. Count development captures reporting behavior; it does not automatically capture severity changes or later payments on already reported claims.
Worked example: There are 180 reported claims and a selected count development factor of 1.10. Estimated ultimate count is 198, implying 18 additional reports under that count definition.
Mistake to avoid: Multiplying unreported counts by a severity estimate that includes a different population of claims.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
56. Severity for outstanding and unreported claims
Outstanding reported claims and claims not yet reported can have different expected costs. Estimate severity for the matching population and distinguish remaining payment severity from total ultimate severity. Already paid amounts must not be included again when calculating the future payment requirement on known claims.
Worked example: Forty outstanding reported claims need $3,000 each in future payments, while ten unreported claims are expected to cost $5,000 each. Estimated unpaid claims are $120,000 + $50,000 = $170,000.
Mistake to avoid: Using total claim severity for outstanding claims when part of their cost has already been paid.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
57. Frequency-severity estimation of ultimate costs
Expected ultimate claim cost equals exposure multiplied by expected claim frequency and expected severity for the same claim definition. This decomposition helps identify whether changes arise from claim incidence or claim size. After estimating ultimate cost, subtract paid claims to obtain unpaid liability on the same coverage and expense basis.
Worked example: With 5,000 exposures, frequency 0.04 and severity $6,000, expected ultimate is $1.2 million. If $450,000 has been paid, estimated unpaid liability is $750,000.
Mistake to avoid: Mixing frequency for all reported claims with severity calculated only from claims receiving payment.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
58. Combining frequency and severity trends
Frequency and severity trends combine multiplicatively when estimating the trend in expected cost per exposure. Opposing movements need not cancel exactly. Keep claim definitions and exposure units stable, and distinguish a change in average severity caused by claim mix from a within-category increase in claim costs.
Worked example: Frequency decreases 3% while severity increases 8%. Expected cost per exposure changes by 0.97 × 1.08 = 1.0476, an increase of 4.76%.
Mistake to avoid: Adding −3% and 8% and reporting an exact 5% cost increase.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
59. Provisioning for unpaid claim-handling expenses
Unpaid claim liabilities may need a separate allowance for future handling expenses not included in projected claims. Choose an expense model reflecting the work remaining, and specify whether expenses are attached to individual claims or shared. Payments still due do not necessarily determine remaining handling work in a fixed proportion.
Worked example: Projected unpaid indemnity is $900,000 and a separately supported remaining handling-expense estimate is $75,000. The combined unpaid provision is $975,000 if those expenses are absent from the indemnity projection.
Mistake to avoid: Adding an expense provision without checking whether the selected claim projection already includes it.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
60. Selecting an estimate from multiple methods
Compare methods on a common paid, expense and recovery basis, then evaluate their assumptions against maturity and operational evidence. A selected estimate should explain why particular methods deserve weight. The spread between estimates is useful diagnostic information, but it is not automatically a probability interval for the eventual outcome.
Worked example: Paid chain ladder estimates $500,000 unpaid and Bornhuetter-Ferguson $420,000. A documented 25% and 75% weighting gives $440,000, provided the assumptions and bases justify that selection.
Mistake to avoid: Calling the $420,000–$500,000 method range a confidence interval without a supporting uncertainty model.
Source reference: Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
Sources
Exam identity verified:
- Exam 5 —Basic Ratemaking and Est Claim Liabilities | Casualty Actuarial Society
- Casualty Actuarial Society