Why Input Consistency Matters
A Cape Cod estimate pools reported losses across periods to estimate an expected loss ratio, then applies that ratio to each period's exposure. The pooling can stabilize estimates, but only if the inputs describe comparable experience. A premium total from one set of periods, losses from another, or development factors measured on a different basis can produce a precise-looking result that does not estimate a meaningful loss ratio.
The key audit question is not just whether the arithmetic is correct. It is whether each numerator and denominator refers to the same periods, loss basis, valuation date, and exposure definition. This tutorial focuses on those consistency checks. The method is aligned with the SOA Exam ASTAM (Advanced Short-Term Actuarial Mathematics) syllabus.
This formulation assumes that the selected exposures are measured consistently across periods and correspond to the loss basis being estimated. It also assumes each \(q_i\) represents the proportion of ultimate losses reported by the valuation date for that period. If premiums are adjusted, the adjustment must be appropriate to the intended loss ratio and applied consistently. The inputs should be identified by period and basis before combining them.
Why the Weighted Denominator Appears
The denominator is not simply total premium, nor is it total premium multiplied by a single average reported proportion. Each period has its own development maturity, so only the portion of its expected ultimate losses corresponding to its reported proportion belongs in the comparison with reported losses.
To see the logic, suppose the expected ultimate loss for each period is represented by a common loss ratio \(\lambda\) applied to its exposure: \(U_i=\lambda P_i\). If proportion \(q_i\) of that ultimate is reported at the valuation date, the corresponding expected reported loss is \(q_iU_i=q_i\lambda P_i\). Summing over periods gives the expected reported total \(\lambda\sum_iP_iq_i\). Equating this to the observed reported total and solving for \(\lambda\) gives the Cape Cod estimate.
This justification also explains why a period-specific check matters: a wrong premium \(P_i\) or reported proportion \(q_i\) changes the amount of expected reported loss attributed to that period. The formula does not correct a mismatch in the input data; it pools that mismatch into the estimated ratio.
Worked Example: Premiums and Losses from Matching Periods
Worked Example: Premiums and Losses from Matching Periods
A hypothetical insurer estimates gross ultimate losses for two accident years. All amounts are in thousands of dollars. Accident year 1 has earned premium of 1,000 and reported losses of 600; its age-to-ultimate development factor is 1.25. Accident year 2 has earned premium of 1,200 and reported losses of 360; its factor is 2.00. All four amounts refer to the same accident years and the same valuation date. Calculate the Cape Cod loss ratio, ultimates, and reserves.
Method. Convert each development factor to a reported proportion, calculate the premium-weighted reported exposure, then apply the resulting common loss ratio to each year's premium.
For accident year 1, \(q_1=1/1.25=0.80\), so its weighted exposure is \(1{,}000(0.80)=800\). For accident year 2, \(q_2=1/2.00=0.50\), so its weighted exposure is \(1{,}200(0.50)=600\).
The reported-loss numerator is \(600+360=960\). The denominator is \(800+600=1{,}400\). Therefore, \(\hat{\lambda}=960/1{,}400=0.685714\), or about 68.57%.
Accident year 1's estimated ultimate is \(1{,}000(0.685714)=685.714\), and its reserve is \(685.714-600=85.714\). Accident year 2's estimated ultimate is \(1{,}200(0.685714)=822.857\), and its reserve is \(822.857-360=462.857\).
The total estimated ultimate is \(685.714+822.857=1{,}508.571\). The total reserve is \(85.714+462.857=548.571\), also equal to \(1{,}508.571-960\). To the nearest 0.01 thousand dollars, the ultimates are 685.71 and 822.86, and the total reserve is 548.57.
Check. The weighted reported exposure is \(1{,}400\), and \(1{,}400(0.685714)=960\), matching total reported losses subject to rounding. The common mistake avoided is using total premium, \(2{,}200\), as the denominator without allowing for the different reported proportions.
Four Compatibility Checks
Before calculating, build a small input table with one row per period. Record the period label, valuation date or age, premium or exposure, reported losses, loss basis, and development factor. This makes mismatches visible before they affect the pooled estimate.
Confirm that each premium and loss amount refers to the same accident or other selected period. Do not pair an accident-year loss total with a calendar-year premium merely because the labels are adjacent.
State whether each quantity is gross or net of reinsurance and whether any adjustments have been applied. The numerator and denominator must support the same intended loss ratio.
Use a factor appropriate to the period's actual development age, the selected paid or incurred loss measure, and the valuation date. Calculate \(q_i=1/C_i\), not \(C_i\), for the reported proportion.
Calculate \(\sum_iP_iq_i\) directly. Do not replace the individual \(q_i\) values with an unweighted average unless that simplification is justified by the exposure pattern and the intended calculation.
Check that \(\hat{\lambda}\sum_iP_iq_i=\sum_iR_i\), subject to rounding, and that total reserve equals total estimated ultimate less total reported losses.
Worked Example: Detecting a Gross-Net Mismatch
Worked Example: Detecting a Gross-Net Mismatch
A hypothetical portfolio has two accident years. Gross earned premiums are 1,000 and 500 thousand dollars. Net earned premiums, after reinsurance, are 800 and 300 thousand dollars. Reported losses are gross: 400 and 150 thousand dollars. The reported proportions are 0.80 and 0.50. Calculate the gross-basis Cape Cod estimate, then show the effect of incorrectly using net premiums with gross reported losses.
Method. For a gross loss estimate, pair gross reported losses with gross premiums. First calculate the correct estimate. Then use the mismatched denominator to quantify why identifying the bases matters.
The gross-basis weighted exposure is \(1{,}000(0.80)+500(0.50)=800+250=1{,}050\). Total gross reported losses are \(400+150=550\). Thus, the gross expected loss ratio is \(550/1{,}050=0.5238095\).
The gross ultimate estimates are \(1{,}000(0.5238095)=523.81\) and \(500(0.5238095)=261.90\). The reserves are \(523.81-400=123.81\) and \(261.90-150=111.90\), for a total of 235.71 thousand dollars. Values are rounded to the nearest 0.01 thousand dollars.
Now use net premiums by mistake. The mismatched weighted exposure is \(800(0.80)+300(0.50)=640+150=790\). This gives the apparent ratio \(550/790=0.6962025\). Applying it to the gross premiums gives apparent ultimates of \(1{,}000(0.6962025)=696.20\) and \(500(0.6962025)=348.10\), and an apparent total reserve of \(696.20+348.10-550=494.30\) thousand dollars.
Check and correction. The mismatched calculation increases the reserve from 235.71 to 494.30 thousand dollars, but that difference is not evidence of a change in claim experience: the inputs describe different reinsurance bases. Use premiums consistent with the gross losses for the gross estimate. If a net estimate is required, use net losses and a clearly defined net exposure basis. The common mistake avoided is treating “premium” as a basis-free quantity.
Worked Example: Development Factors and Exposure Weights
Worked Example: Development Factors and Exposure Weights
A hypothetical insurer is estimating losses for three accident years, with all premiums and reported losses in thousands of dollars. The premiums are 1,000, 600, and 400; the reported losses are 500, 240, and 80. At the same valuation date, the age-to-ultimate factors are 1.25, 2.00, and 4.00, respectively. Calculate the Cape Cod estimate and compare it with a calculation that incorrectly substitutes the unweighted average reported proportion for the three period-specific proportions.
Method. Invert each factor separately, form each period's weighted exposure, and pool the reported losses over their sum.
The reported proportions are \(q_1=1/1.25=0.80\), \(q_2=1/2.00=0.50\), and \(q_3=1/4.00=0.25\). The corresponding weighted exposures are \(1{,}000(0.80)=800\), \(600(0.50)=300\), and \(400(0.25)=100\). Their sum is \(1{,}200\).
Total reported losses are \(500+240+80=820\). Therefore, \(\hat{\lambda}=820/1{,}200=0.683333\), or about 68.33%. The estimated ultimates are \(1{,}000(0.683333)=683.33\), \(600(0.683333)=410.00\), and \(400(0.683333)=273.33\). The corresponding reserves are \(183.33\), \(170.00\), and \(193.33\), totaling \(546.67\) thousand dollars.
For comparison, the unweighted average reported proportion is \((0.80+0.50+0.25)/3=0.516667\). Multiplying total premium of 2,000 by this average gives an incorrect denominator of \(1{,}033.333\). The resulting apparent ratio is \(820/1{,}033.333=0.793548\). Applied to total premium, this implies an apparent total ultimate of \(2{,}000(0.793548)=1{,}587.10\) and reserve of \(1{,}587.10-820=767.10\) thousand dollars.
Check and correction. The correct denominator is \(800+300+100=1{,}200\), not \(2{,}000\) times the unweighted average proportion. Because the periods have different premiums, each period's proportion must be weighted by its own premium. The common mistake avoided is averaging development proportions as if every period had equal exposure.
Common Exam Traps and Final Audit
A frequent error is to confuse a cumulative development factor with a reported proportion. If \(C_i=2\), the estimated reported proportion is \(1/2=0.50\), not 2.00 and not \(1-1/2\). The complement, \(1-q_i\), is the estimated unreported proportion; it is not the weight used in the Cape Cod expected loss ratio denominator.
Another trap is a period mismatch hidden by similar labels. A premium schedule may be organized by calendar year while a loss triangle is organized by accident year. The values cannot be paired row by row without establishing that they represent compatible exposure and losses. Likewise, a factor selected for a different valuation date or for paid losses cannot automatically be applied to incurred reported losses.
Units and adjustments also need attention. If premiums are in thousands but losses are in dollars, convert one set before calculating. If premium has been trended or otherwise adjusted, state which premium definition is being used and ensure it represents the exposure basis intended for the loss ratio. A numerically valid ratio is not meaningful if the numerator and denominator use different definitions.
The final check should show the period-level inputs and the calculations, not merely a reported total. Confirm that the pooled numerator is the sum of the specified reported losses, the denominator is the sum of period-specific exposure times reported proportion, and the reserve reconciles to ultimate less reported. A mismatch should be resolved in the data definition, not hidden by adjusting the final answer.
Check Your Understanding
For each problem, identify the compatible inputs before calculating. Dollar amounts are in thousands unless stated otherwise.
- A hypothetical accident year has premium 900, reported losses 360, and an age-to-ultimate factor of 1.25. Calculate its Cape Cod loss ratio, ultimate, and reserve. (Answer: \(q=0.80\), loss ratio \(0.5000\), ultimate 450, reserve 90.)
- Two compatible periods have premiums 800 and 500, reported losses 300 and 150, and age-to-ultimate factors 1.60 and 2.50. Calculate the weighted denominator, loss ratio, and total reserve. (Answer: denominator 700, loss ratio \(0.6429\), total reserve 385.71.)
- A hypothetical portfolio's reported losses are gross of reinsurance, but the supplied premiums are net of reinsurance. Is the resulting ratio a valid gross loss ratio? State the required correction. (Answer: No; use premiums consistent with gross losses for a gross estimate, or use compatible net losses and net premiums for a net estimate.)
- For one period, premium is 700, reported losses are 200, and the age-to-ultimate factor is 2.00. A calculation uses 2.00 as the reported proportion. Give the correct proportion, loss ratio, ultimate, and reserve. (Answer: \(q=0.50\), loss ratio \(0.5714\), ultimate 400, reserve 200.)
- Two periods have premiums 1,000 and 400, reported losses 400 and 80, and reported proportions 0.80 and 0.25. Calculate the correct pooled loss ratio and explain why the unweighted average of the proportions is unsuitable. (Answer: denominator \(1{,}000(0.80)+400(0.25)=900\), loss ratio \(480/900=0.5333\); the proportions require premium weighting because the period exposures differ.)