Deriving the Cape Cod Loss Ratio

Introduction

The Cape Cod method uses observed loss experience to calibrate an expected loss ratio across accident periods. Its central calculation is a ratio of two aggregates: reported losses in the numerator, and exposure adjusted for how much of each period's ultimate loss is expected to have been reported in the denominator. The denominator matters because a recent period has had less time to report losses than a mature one.

The derivation is short, but the inputs must be consistent. Reported losses, exposure, and the selected reported proportion must refer to the same periods and a compatible basis. This tutorial derives the estimate and shows how to check those conditions. It is aligned with the SOA Exam ASTAM (Advanced Short-Term Actuarial Mathematics) syllabus.

Definition: For accident period $j$, let $C_j$ be cumulative reported losses at the valuation date, $E_j$ be the exposure measure used to scale expected ultimate losses, and $q_j$ be the selected proportion of ultimate losses expected to be reported by that date. Under a common expected loss ratio $R$, the expected ultimate losses for period $j$ are $E_jR$, and the expected reported losses are $E_jRq_j$. The Cape Cod expected loss ratio is $$ \widehat{R}_{\mathrm{CC}}= \frac{\sum_j C_j}{\sum_j E_jq_j}, $$ provided the denominator is positive.

Why the Denominator Is Exposure-Weighted

The expected loss ratio $R$ converts exposure into expected ultimate losses. But the numerator contains reported losses, not ultimate losses. Therefore, each period's expected ultimate losses must be multiplied by its reported proportion before it can be compared with reported experience. The quantity $E_jq_j$ is the period's exposure adjusted for its reporting maturity.

For example, if two periods have equal exposure but one is expected to have reported 80% of its ultimate losses while the other has reported 40%, their expected reported losses are not equal under the same expected loss ratio. The more mature period contributes twice as much exposure to the denominator. The method does not simply divide reported losses by total premium or average period-level loss ratios.

Suppose a set of accident periods shares a common expected loss ratio $R$. For period $j$, expected ultimate losses are $E_jR$. If the selected reported proportion is $q_j$, the expected reported amount is $E_jRq_j$. Adding across periods gives the aggregate expected reported amount. The Cape Cod calibration equates that amount to the aggregate observed reported losses.

State expected ultimate losses.
For period $j$, the common loss ratio gives expected ultimate losses $E_jR$.

Convert to expected reported losses.
Only the proportion $q_j$ is expected to be reported at the valuation date, so expected reported losses are $E_jRq_j$.

Aggregate and equate.
Sum expected reported losses and set that total equal to observed reported losses.

Solve for the common ratio.
Divide by total exposure adjusted for reporting maturity, provided that total is positive.

\sum_j C_j = \sum_j E_jRq_j = R\sum_j E_jq_j \quad\Longrightarrow\quad \widehat{R}_{\mathrm{CC}} = \frac{\sum_j C_j}{\sum_j E_jq_j}.

This is an aggregate calibration: it chooses the common ratio that makes total expected reported losses equal total observed reported losses. It does not require the expected reported amount to equal the observed amount separately in every period. The formula is not a probability model for the observed totals; it is an actuarial estimate based on the stated common-ratio and reporting assumptions.

Required inputs and conditions: The calculation requires (1) reported losses $C_j$ for each included period, on a consistent reported-loss basis; (2) an exposure measure $E_j$ that scales expected ultimate losses on a matching basis; and (3) a selected reported proportion $q_j$ for each period at the same valuation date. The aggregate denominator $\sum_j E_jq_j$ must be positive. The common-ratio interpretation also assumes that one expected loss ratio is appropriate across the included periods after exposure and reporting maturity are accounted for.

Obtaining and Checking the Reported Proportion

The reported proportion is the expected fraction of ultimate losses already reported at the valuation date. When development is expressed as a cumulative reported-loss development factor $L_j$, defined as ultimate losses divided by reported losses at the period's current maturity, the corresponding reported proportion is

q_j=\frac{1}{L_j}.

This conversion applies when the selected factor and the reported losses use the same loss basis and valuation maturity. A factor of 2.0, for instance, means ultimate losses are expected to be twice current reported losses, so the reported proportion is $1/2=0.50$. If a selected development input is already a reported proportion, use that proportion directly; do not take its reciprocal. For a paid-loss calculation, the development information must describe paid losses, not reported incurred losses.

A reported proportion should ordinarily be between 0 and 1. A value of 1 represents a period treated as fully reported; a value of 0 implies no expected reported losses at that valuation date. The latter creates a special issue in the pooled formula, discussed below. Development selections should also be consistent with the valuation date and should include any applicable tail assumption if the selected ultimate factor requires one.

Worked Examples

Worked Example: Two Periods With Different Reporting Maturities

A hypothetical portfolio has two accident periods. Period 1 has earned premium of $300,000, a selected reported proportion of 0.40, and cumulative reported losses of $90,000. Period 2 has earned premium of $500,000, a selected reported proportion of 0.75, and cumulative reported losses of $210,000. Using earned premium as the exposure measure, calculate the Cape Cod expected loss ratio.

Method. Apply the aggregate Cape Cod formula. The exposures are earned premiums, and the reported proportions are supplied directly.

Calculate the adjusted exposure. For period 1, $E_1q_1=\$300{,}000(0.40)=\$120{,}000$. For period 2, $E_2q_2=\$500{,}000(0.75)=\$375{,}000$. Thus, the denominator is $\$120{,}000+\$375{,}000=\$495{,}000$.

Aggregate reported losses. The numerator is $\$90{,}000+\$210{,}000=\$300{,}000$.

Calculate the ratio. Therefore, $$ \widehat{R}_{\mathrm{CC}} = \frac{\$300{,}000}{\$495{,}000} = 0.6060606\ldots $$ As a percentage, the estimate is 60.61%, rounded to two decimal places.

Check. At a ratio of approximately 60.61%, expected reported losses are approximately $\$120{,}000(0.6061)=\$72{,}732$ and $\$375{,}000(0.6061)=\$227{,}288$. Their sum is approximately $300,020, with the small difference from $300,000 due to rounding; using the unrounded ratio gives exactly $300,000.

Worked Example: Why Period Ratios Must Be Weighted

A hypothetical book has two periods. Period 1 has exposure $E_1=\$100{,}000$, reported proportion $q_1=0.40$, and reported losses $C_1=\$50{,}000$. Period 2 has exposure $E_2=\$300{,}000$, reported proportion $q_2=0.80$, and reported losses $C_2=\$180{,}000$. Calculate the Cape Cod ratio and compare it with the unweighted average of the two period-level ratios.

Method. First use aggregate reported losses divided by aggregate exposure-weighted reported proportions. Then express that result as a weighted average to see why a simple average is inappropriate.

Calculate adjusted exposures and totals. Period 1 contributes $\$100{,}000(0.40)=\$40{,}000$. Period 2 contributes $\$300{,}000(0.80)=\$240{,}000$. The denominator is $\$40{,}000+\$240{,}000=\$280{,}000$, and total reported losses are $\$50{,}000+\$180{,}000=\$230{,}000$.

Calculate the aggregate estimate. $$ \widehat{R}_{\mathrm{CC}} = \frac{\$230{,}000}{\$280{,}000} = 0.8214286\ldots = 82.14\%. $$ The percentage is rounded to two decimal places.

Compare with the period-level ratios. Period 1's ratio is $\$50{,}000/\$40{,}000=1.25$. Period 2's ratio is $\$180{,}000/\$240{,}000=0.75$. Their unweighted average is $(1.25+0.75)/2=1.00$, or 100.00%, not the Cape Cod estimate.

The correct weights are the adjusted exposures: $\$40{,}000$ and $\$240{,}000$. Weighting the period ratios gives $$ \frac{\$40{,}000(1.25)+\$240{,}000(0.75)} {\$40{,}000+\$240{,}000} = \frac{\$50{,}000+\$180{,}000}{\$280{,}000} = 0.8214286\ldots. $$ Check. The weighted expression agrees with the aggregate formula. The common mistake is to give a small adjusted-exposure period the same influence as a much larger one.

Worked Example: Converting Development Factors to Reported Proportions

A hypothetical liability portfolio has two accident periods. Each has earned premium of $200,000. The selected cumulative reported-loss development factors to ultimate are 2.00 for Period A and 1.25 for Period B. Reported losses are $60,000 for Period A and $168,000 for Period B. Calculate the Cape Cod expected loss ratio using the factors to derive reported proportions.

Method. Because each factor is ultimate divided by current reported losses, convert each factor to a reported proportion by taking its reciprocal.

Convert the factors. For Period A, $q_A=1/2.00=0.50$. For Period B, $q_B=1/1.25=0.80$.

Calculate the denominator. Period A contributes $\$200{,}000(0.50)=\$100{,}000$. Period B contributes $\$200{,}000(0.80)=\$160{,}000$. Total adjusted exposure is $\$100{,}000+\$160{,}000=\$260{,}000$.

Calculate the numerator and ratio. Total reported losses are $\$60{,}000+\$168{,}000=\$228{,}000$. Hence, $$ \widehat{R}_{\mathrm{CC}} = \frac{\$228{,}000}{\$260{,}000} = 0.8769231\ldots = 87.69\%. $$ The percentage is rounded to two decimal places.

Check. Period A's current reported losses are 50% of its indicated ultimate amount under the selected factor, while Period B's are 80%. The denominator therefore assigns more adjusted exposure to Period B. The common trap is to use 2.00 and 1.25 as reported proportions; they are development factors, so the required proportions are their reciprocals.

Weighted-Ratio Identity and Zero-Weight Periods

When every included period has positive adjusted exposure $E_jq_j$, the aggregate estimate can also be written as a weighted average of period-level ratios. The algebra follows by multiplying and dividing each reported amount by its adjusted exposure:

\frac{\sum_j C_j}{\sum_j E_jq_j} = \frac{\sum_j (E_jq_j)\left(C_j/(E_jq_j)\right)} {\sum_j E_jq_j}.

This identity requires $E_jq_j>0$ for every period included in the period-ratio sum, so that $C_j/(E_jq_j)$ is defined. If a period has zero adjusted exposure and also has zero reported losses, it can be omitted from both sums without changing the estimate. If a zero-adjusted-exposure period has positive reported losses, its period-level ratio is undefined and it cannot be treated as an ordinary weighted-average component. The pooled quotient may still be arithmetically calculable if other periods make the denominator positive, but that zero-weight period's reported losses cannot be explained by the common-ratio expected-reported model. Investigate the data or the selected inputs before interpreting the result.

Common Exam Traps and Practical Checks

Using total exposure without maturity adjustment. The denominator is $\sum_j E_jq_j$, not $\sum_j E_j$, unless every included period has $q_j=1$.
Using the development factor as the reported proportion. If the factor is ultimate divided by current reported losses, the reported proportion is its reciprocal.
Averaging period ratios without weights. When every adjusted exposure is positive, the relevant weights are $E_jq_j$, not equal weights and not exposure alone.
Mixing data bases. Premium and reported losses should be aligned to the same business and period definitions. Development selections must match the loss basis and valuation maturity of the reported amounts.
Ignoring a zero-weight period with reported losses. A positive $C_j$ when $E_jq_j=0$ signals a problem for the period-level model, even if the pooled quotient can be computed.

A useful final check is dimensional: if $E_j$ is measured in dollars of premium and $q_j$ is dimensionless, then $\sum_j E_jq_j$ is in dollars of premium, and the resulting ratio has units of losses per dollar of premium. Also verify that the reported losses in the numerator are cumulative amounts at the same valuation date used to select the reported proportions.

Key takeaway: The Cape Cod expected loss ratio is aggregate reported losses divided by aggregate exposure adjusted for reporting maturity. Its derivation depends on a common loss-ratio assumption, consistent loss and exposure bases, appropriate reported proportions, and a positive adjusted-exposure denominator.

Check Your Understanding

For each problem, use the aggregate Cape Cod formula and check that the exposure and reported proportion are on compatible bases.

A hypothetical accident period has earned premium of $250,000, a reported proportion of 0.60, and reported losses of $105,000. What is the Cape Cod expected loss ratio? (Answer: 70.00%)
Two hypothetical periods have exposures of $100,000 and $200,000, reported proportions of 0.50 and 0.75, and reported losses of $45,000 and $105,000, respectively. Calculate the aggregate Cape Cod ratio. (Answer: 75.00%)
A hypothetical period has exposure of $400,000, reported losses of $80,000, and a cumulative reported-loss development factor of 4.00 to ultimate. Convert the factor to a reported proportion, then calculate the Cape Cod ratio. (Answer: 80.00%)
Two hypothetical periods each have exposure of $200,000. Their cumulative reported-loss development factors are 1.25 and 2.50, and their reported losses are $130,000 and $90,000. Calculate the Cape Cod ratio. (Answer: 91.67%)
A hypothetical portfolio has Period 1 exposure of $100,000, reported proportion 0.50, and reported losses of $40,000. Period 2 has zero exposure, reported proportion 0.60, and reported losses of $10,000. What quotient does the aggregate formula produce, and why should it not be interpreted as a valid common-ratio fit to both periods? (Answer: 100.00%; Period 2 has zero adjusted exposure but positive reported losses, so its period-level expected-reported amount cannot be explained by the common-ratio model.)

Deriving the Cape Cod Loss Ratio

What You'll Learn

Introduction

Why the Denominator Is Exposure-Weighted