Deriving the Cape Cod Expected Loss Ratio

Why the Cape Cod Ratio Uses Emerged Premium

For an immature accident year, observed losses are only a portion of the losses expected to emerge by ultimate. The Cape Cod method estimates a common expected loss ratio from the observed experience across accident years while recognizing that each year is at a different stage of development. The key adjustment is to count only the share of each year’s earned premium corresponding to the proportion of ultimate losses expected to have emerged.

This tutorial derives that adjustment. It focuses on the aggregate expected loss ratio calculation, not on allocating the resulting expected losses to individual years or estimating ultimate losses and reserves.

Definition: For accident year $i$, let $EP_i$ be earned premium on the selected, consistent premium basis; let $L_i$ be observed losses at the evaluation date on a consistent paid or incurred basis; and let $w_i$ be the proportion of ultimate losses expected to have emerged by that date. The Cape Cod expected loss ratio estimate, denoted $R_{CC}$, is $$ R_{CC}=\frac{\sum_i L_i}{\sum_i EP_i w_i}. $$ The denominator is the total earned premium adjusted by each accident year’s emergence proportion. The $w_i$ values must be expressed as proportions between zero and one, not as percentages such as 80.

Here, emerged means reflected in the chosen observed-loss measure by the evaluation date. For example, if the analysis uses paid losses, the emergence proportions must describe the expected proportion of ultimate losses paid. If it uses incurred losses, they must be appropriate for incurred losses. Premium, loss, and emergence assumptions must refer to a consistent set of accident years and valuation conditions.

Deriving the Estimator

Assume that the accident years share a common expected ultimate loss ratio $R$, and that the selected $w_i$ values represent the proportions of ultimate losses expected to be observed for the corresponding years. The expected ultimate losses for year $i$ are then $EP_iR$. Under the emergence assumption, the expected observed losses are $EP_iR w_i$.

Summing those expected observed losses over the years gives the model’s expected aggregate observed losses. The observed aggregate is used to estimate that expectation:

E\left[\sum_i L_i\right] =\sum_i EP_i R w_i =R\sum_i EP_i w_i.

Solving this aggregate relationship for $R$, with the observed aggregate losses substituted for their expectation, yields the Cape Cod estimator:

\widehat{R}_{CC} =\frac{\sum_i L_i}{\sum_i EP_i w_i}.

This is an aggregate method-of-moments justification: the selected ratio makes the model’s expected aggregate observed losses equal the actual aggregate observed losses. It does not assert that every accident year’s observed losses equal its expectation. Random variation by year is compatible with the aggregate calculation.

Choose consistent inputs.
Identify the accident years, earned premiums, observed-loss basis, and evaluation date. Select emergence proportions appropriate to that loss basis and date.

Calculate adjusted premium by year.
For each year, multiply earned premium by its emergence proportion: $EP_iw_i$.

Aggregate and divide.
Sum observed losses and divide by the sum of adjusted premiums. The result is a loss ratio, so report it as a percentage or decimal with units understood.

Worked Calculations

Worked Example: Deriving a Ratio Across Three Maturity Levels

A hypothetical insurer reviews three accident years. Earned premiums are $2.00 million, $3.00 million, and $4.00 million. The selected proportions of ultimate incurred losses expected to have emerged are 80%, 50%, and 25%, respectively. Observed incurred losses are $1.20 million, $1.00 million, and $0.40 million. Estimate the common Cape Cod expected loss ratio.

Method and common mistake avoided. Use total observed incurred losses divided by total earned premium weighted by the incurred emergence proportions. Do not divide by total unadjusted premium: that would treat every year as fully emerged.

First, convert each percentage to a proportion and calculate adjusted premium:

Year 1: $EP_1w_1=\$2.00\text{ million}\times0.80=\$1.60\text{ million}$.
Year 2: $EP_2w_2=\$3.00\text{ million}\times0.50=\$1.50\text{ million}$.
Year 3: $EP_3w_3=\$4.00\text{ million}\times0.25=\$1.00\text{ million}$.

The adjusted-premium denominator is $\$1.60+\$1.50+\$1.00=\$4.10$ million. The observed-loss numerator is $\$1.20+\$1.00+\$0.40=\$2.60$ million. Therefore,

\widehat{R}_{CC} =\frac{\$2.60\text{ million}}{\$4.10\text{ million}} =0.634146\ldots

Rounded to two decimal places as a percentage, the estimate is 63.41%. Check: multiplying the denominator by the unrounded ratio gives $\$4.10\text{ million}\times0.634146\ldots=\$2.60\text{ million}$, the observed aggregate loss.

Worked Example: Keeping the Loss Basis and Emergence Basis Aligned

Consider a hypothetical portfolio with earned premiums of $1.20 million, $2.40 million, and $3.00 million for three accident years. At the evaluation date, the observed losses selected for this calculation are paid losses of $0.72 million, $0.84 million, and $0.45 million. The corresponding proportions of ultimate paid losses expected to have emerged are 90%, 60%, and 30%. Calculate the Cape Cod ratio.

Method and common mistake avoided. Apply the paid-loss emergence proportions because the numerator is paid losses. The method would not be consistent if these proportions described incurred emergence instead.

The adjusted premiums are:

Year 1: $\$1.20\text{ million}\times0.90=\$1.08\text{ million}$.
Year 2: $\$2.40\text{ million}\times0.60=\$1.44\text{ million}$.
Year 3: $\$3.00\text{ million}\times0.30=\$0.90\text{ million}$.

Thus, the denominator is $\$1.08+\$1.44+\$0.90=\$3.42$ million. The numerator is $\$0.72+\$0.84+\$0.45=\$2.01$ million. The estimate is

\widehat{R}_{CC} =\frac{\$2.01\text{ million}}{\$3.42\text{ million}} =0.587719\ldots

Rounded to two decimal places as a percentage, the result is 58.77%. A useful arithmetic check is $\$3.42\text{ million}\times0.587719\ldots=\$2.01\text{ million}$. The ratio is dimensionless because both numerator and denominator are dollar amounts.

Worked Example: Quantifying the Unweighted-Premium Error

A hypothetical program has three accident years, with earned premiums of $1.00 million, $2.00 million, and $1.00 million. The proportions of ultimate losses expected to have emerged are 90%, 50%, and 20%. Observed losses total $0.90 million. Calculate the Cape Cod ratio and compare it with the result from incorrectly using full earned premium.

Method and common mistake avoided. Weight each year’s premium by its emergence proportion before summing. The deliberate incorrect calculation below shows the bias introduced by ignoring maturity.

The adjusted premiums are $\$1.00\text{ million}\times0.90=\$0.90$ million, $\$2.00\text{ million}\times0.50=\$1.00$ million, and $\$1.00\text{ million}\times0.20=\$0.20$ million. Their sum is $\$2.10$ million. Therefore, the correct estimate is

\widehat{R}_{CC} =\frac{\$0.90\text{ million}}{\$2.10\text{ million}} =0.428571\ldots =42.86\%.

The incorrect unweighted calculation is

\frac{\$0.90\text{ million}} {\$1.00\text{ million}+\$2.00\text{ million}+\$1.00\text{ million}} =\frac{0.90}{4.00} =22.50\%.

That calculation assumes all $4.00 million of premium corresponds to losses already emerged, even though the selected proportions imply only $2.10 million of adjusted premium. The corrected answer is 42.86%, not 22.50%. The difference is not a rounding issue; it comes from omitting the emergence adjustment.

Assumptions and Interpretation

The derivation is useful only when its inputs support the model. In particular, it assumes a common expected ultimate loss ratio across the accident years being pooled. If expected loss ratios differ materially by year, a single aggregate ratio may not describe the experience appropriately. The method also relies on the selected emergence proportions being suitable estimates for the observed-loss definition and evaluation date.

The denominator is sometimes called emerged premium or premium adjusted for emergence. It is a mathematical device for relating observed losses to an expected ultimate loss ratio; it is not a separate amount of premium actually earned. A year with more earned premium contributes more to the denominator, while a year with a smaller emerged proportion contributes less, all else equal.

Input or condition	Required interpretation	What can go wrong
Earned premium $EP_i$	Premium for the same accident year and selected portfolio basis as the loss analysis	Mixing premium periods or bases makes the loss ratio difficult to interpret
Observed losses $L_i$	Losses observed by the evaluation date, consistently measured across years	Mixing paid and incurred amounts changes the numerator’s meaning
Emergence proportion $w_i$	Expected share of ultimate losses observable on the chosen basis and date	Using the wrong maturity or loss basis misstates adjusted premium
Common ratio assumption	A single expected ultimate loss ratio is reasonable for the years pooled	Substantial differences in risk mix or conditions may make the aggregate ratio unrepresentative

A proportion must be between zero and one. If all $EP_iw_i$ values sum to zero, the estimator is undefined because there is no adjusted-premium base. If the denominator is positive but very small, the estimate can be highly sensitive to the observed losses. An unusual or negative estimate should prompt a review of the data, definitions, and assumptions rather than an automatic interpretation as a credible expected ratio.

Exam Traps and a Practical Check

Using the percentage as a whole number. If emergence is 75%, use $w_i=0.75$, not $75$. The latter inflates adjusted premium by a factor of 100.
Weighting losses instead of premium in the estimator. The observed losses are summed as given. The emergence adjustment belongs in the premium denominator because the model says expected observed losses equal $EP_i Rw_i$.
Using full premium for immature years. This treats unobserved expected losses as if they had already emerged and generally makes the denominator too large.
Pairing paid losses with incurred emergence, or the reverse. The selected emergence proportion must match the numerator’s loss basis.
Mixing accident-year populations or valuation dates. Each input must refer to a compatible set of years and the same evaluation context.

The central check follows directly from the derivation: after computing the ratio, multiply it by the total adjusted premium. The product should equal the total observed losses, apart from rounding:

\widehat{R}_{CC}\left(\sum_i EP_iw_i\right) =\sum_i L_i.

This identity checks the arithmetic, not the appropriateness of the assumptions. A correct calculation can still be a poor estimate if the emergence proportions are unsuitable or the common-ratio assumption is unreasonable. The derivation is aligned with the CAS Exam 5 (Basic Ratemaking and Estimated Claim Liabilities) syllabus.

Check Your Understanding

For each problem, use observed losses divided by the sum of earned premiums weighted by their emergence proportions. Report ratios to four decimal places unless a percentage is requested.

A hypothetical insurer has earned premiums of $800,000 and $1,200,000, emergence proportions of 75% and 50%, and observed losses of $420,000 and $330,000. What is the Cape Cod ratio? (Answer: 62.50%)
Two accident years have earned premiums of $1.50 million and $2.00 million, emergence proportions of 80% and 30%, and observed losses of $0.90 million and $0.50 million. Calculate the ratio. (Answer: 77.78%)
For three years, earned premiums are $500,000, $700,000, and $800,000; emergence proportions are 90%, 60%, and 25%; and observed losses are $300,000, $250,000, and $150,000. Find the ratio. (Answer: 65.42%)
Observed losses total $450,000. Earned premiums are $600,000, $900,000, and $500,000, with corresponding emergence proportions of 75%, 50%, and 25%. Calculate the ratio and identify the adjusted-premium denominator. (Answer: denominator $1,025,000; ratio 43.90%)
A selected common ratio is 48%. Three years have earned premiums of $1.00 million, $2.00 million, and $3.00 million, with emergence proportions of 90%, 50%, and 20%. Observed losses in the first two years are $350,000 and $500,000. What observed loss amount in the third year would make the aggregate data consistent with the selected ratio? (Answer: $350,000)

Input or condition	Required interpretation	What can go wrong
Earned premium \(EP_i\)	Premium for the same accident year and selected portfolio basis as the loss analysis	Mixing premium periods or bases makes the loss ratio difficult to interpret
Observed losses \(L_i\)	Losses observed by the evaluation date, consistently measured across years	Mixing paid and incurred amounts changes the numerator’s meaning
Emergence proportion \(w_i\)	Expected share of ultimate losses observable on the chosen basis and date	Using the wrong maturity or loss basis misstates adjusted premium
Common ratio assumption	A single expected ultimate loss ratio is reasonable for the years pooled	Substantial differences in risk mix or conditions may make the aggregate ratio unrepresentative

What You'll Learn