The Cape Cod Method as an Experience-Based Prior

Why Pool Experience Across Accident Periods?

An immature accident period has not yet produced all the losses that will ultimately be paid or reported. Its current experience alone can therefore give a misleading picture of its eventual loss level. Bornhuetter-Ferguson addresses this by using an expected loss ratio as a baseline for the unreported portion. The Cape Cod method makes that baseline experience-based: it estimates a common loss ratio from the reported losses and emerged exposure across several accident periods.

The key is to allow for maturity when pooling. A period that is only partly developed contributes less emerged exposure than a period that is nearly or fully developed. Cape Cod therefore does not simply divide total reported losses by total premium. It divides by exposure adjusted for the proportion of losses expected to have emerged.

Definition: For accident period $i$, let $E_i$ be its exposure measure (often earned premium), $L_i$ its cumulative loss amount on the valuation basis, and $p_i$ the proportion of ultimate losses expected to have emerged on that same basis. The Cape Cod common expected loss ratio is $$ \widehat{r}_{CC}=\frac{\sum_i L_i}{\sum_i E_i p_i}. $$ The estimated unreported loss for period $i$ is $$ \widehat{U}_i=E_i\widehat{r}_{CC}(1-p_i). $$

The loss amount $L_i$ and the proportion $p_i$ must refer to the same basis. For paid losses, use cumulative paid losses and the expected paid proportion; for incurred losses, use cumulative incurred losses and the expected incurred proportion. The exposure measure must also be suitable for the loss ratio being estimated. These calculations concern loss development, not life-contingency cash flows: interest and mortality assumptions are not used.

How the Experience-Based Estimate Works

The ratio $\widehat{r}_{CC}$ is common across the periods in the calculation. The denominator $\sum_i E_i p_i$ is the total exposure adjusted for how much loss emergence is expected by the valuation date. Thus a period with exposure of $E_i$ and emergence proportion $p_i$ contributes $E_i p_i$ to the denominator, rather than its full exposure $E_i$.

Once the common ratio is estimated, the expected unreported amount for a particular period depends on that period’s own exposure and unreported proportion $1-p_i$. A mature period with $p_i=1$ contributes fully to the denominator but has no expected unreported amount under this calculation. An immature period has a larger unreported proportion and receives a larger expected provision, all else equal.

Interpretation: Cape Cod uses aggregate experience to set a common expected loss ratio, then applies that ratio to each period’s exposure that has not yet emerged. The resulting $\widehat{U}_i$ is estimated unreported loss; it is not the period’s cumulative reported loss.

A short moment-based argument explains the maturity adjustment. If the underlying expected loss ratio is $r$, then expected ultimate losses for period $i$ are $E_i r$. If the expected emerged proportion is $p_i$, expected cumulative losses at the valuation date are $E_i r p_i$. Adding those expected cumulative amounts across periods gives an expected total of $r\sum_i E_i p_i$. Equating that expected total to the observed aggregate $\sum_i L_i$ gives the Cape Cod estimate above. This justification relies on a common loss ratio and on the selected emergence proportions being appropriate.

Put the data on a consistent basis.
For each accident period, identify its exposure, cumulative loss amount, and expected emerged proportion at the valuation date.

Adjust exposure for emergence.
Calculate $E_i p_i$ for every period and sum these adjusted exposures.

Estimate the common ratio.
Divide total cumulative losses by total adjusted exposure.

Estimate unreported losses by period.
Multiply each period’s exposure by the common ratio and its own unreported proportion $1-p_i$.

Worked Calculations

Worked Example: Pooling Periods at Different Maturities

A hypothetical insurer reviews three accident periods using incurred losses and earned premium, both in thousands of dollars. At the valuation date, the data are:

Period	Earned premium $E_i$	Incurred losses $L_i$	Incurred proportion emerged $p_i$
1	1,000	300	0.500
2	2,000	900	0.750
3	3,000	1,200	1.000

Estimate the common Cape Cod loss ratio and the unreported losses for each period. No interest or mortality assumption is relevant; the data and emergence proportions are all on an incurred basis.

Method. Adjust each period’s premium by its incurred proportion emerged, pool the reported incurred losses, and apply the resulting ratio to each period’s unreported proportion.

The adjusted exposures, in thousands, are $1{,}000(0.500)=500$, $2{,}000(0.750)=1{,}500$, and $3{,}000(1.000)=3{,}000$. Their sum is $500+1{,}500+3{,}000=5{,}000$. The total incurred losses are $300+900+1{,}200=2{,}400$. Therefore,

\widehat{r}_{CC}=\frac{2{,}400}{5{,}000}=0.48=48.00\%.

For period 1, the unreported proportion is $1-0.500=0.500$, so $1{,}000(0.48)(0.500)=240$ thousand dollars. For period 2, it is $1-0.750=0.250$, so $2{,}000(0.48)(0.250)=240$ thousand dollars. For period 3, it is $1-1.000=0$, so the estimate is $3{,}000(0.48)(0)=0$. The estimated unreported total is $240+240+0=480$ thousand dollars.

As a check, the estimated ultimate for each period is its incurred amount plus its estimated unreported amount: $300+240=540$, $900+240=1{,}140$, and $1{,}200+0=1{,}200$, all in thousands. The aggregate is $2{,}880$, which also equals the observed $2{,}400$ plus the estimated unreported $480$. The common mistake avoided here is to call $E_i\widehat{r}_{CC}$ the estimated unreported amount; that is a full-exposure expected loss amount, not the unreported portion.

Worked Example: A Fully Mature Period Still Informs the Ratio

A hypothetical portfolio has three accident periods, each with earned premium of $1{,}000$ thousand dollars. Cumulative paid losses are $180$, $600$, and $870$ thousand dollars, and the expected paid proportions are $0.400$, $0.800$, and $1.000$, respectively. Estimate the common ratio and each period’s unreported paid losses. Use the paid basis throughout; no interest or mortality assumption applies.

Method. Pool paid losses over premium adjusted by the expected paid proportion. Then apply the inferred ratio to each period’s unreported proportion.

The adjusted exposures are $1{,}000(0.400)=400$, $1{,}000(0.800)=800$, and $1{,}000(1.000)=1{,}000$ thousand dollars. Their sum is $2{,}200$. Total paid losses are $180+600+870=1{,}650$ thousand dollars. Hence

\widehat{r}_{CC}=\frac{1{,}650}{2{,}200}=0.75=75.00\%.

The estimated unreported paid losses are:

Period 1: $1{,}000(0.75)(1-0.400)=450$ thousand dollars.
Period 2: $1{,}000(0.75)(1-0.800)=150$ thousand dollars.
Period 3: $1{,}000(0.75)(1-1.000)=0$ thousand dollars.

The total is $450+150+0=600$ thousand dollars. The fully mature third period still contributes $1{,}000$ thousand dollars to the adjusted-exposure denominator; it simply has no unreported proportion to which the ratio can be applied. A common mistake is to omit its exposure from the ratio calculation because its remaining estimate is zero. That would incorrectly change the common ratio. The check is that estimated unreported loss is nonnegative for every period and is zero for the period with $p_i=1$.

Worked Example: Why the Pooled Ratio Is Not an Unweighted Average

A hypothetical insurer has three accident periods. Earned premiums are $500$, $1{,}500$, and $2{,}000$ thousand dollars; the respective emergence proportions are $0.400$, $0.800$, and $1.000$; and cumulative incurred losses are $120$, $840$, and $1{,}300$ thousand dollars. Calculate the Cape Cod ratio and the estimated unreported losses. All amounts and emergence proportions use the incurred basis.

Method. Use the ratio of pooled losses to pooled adjusted exposure. Do not give each period’s individual loss-ratio indication equal weight unless the data happen to imply those weights.

The adjusted exposures are $500(0.400)=200$, $1{,}500(0.800)=1{,}200$, and $2{,}000(1.000)=2{,}000$ thousand dollars, for a total of $3{,}400$. The losses total $120+840+1{,}300=2{,}260$ thousand dollars. Thus

\widehat{r}_{CC}=\frac{2{,}260}{3{,}400}=0.66470588\ldots=66.47\%.

The unreported amounts, using the unrounded ratio in the calculations and rounding money to the nearest $0.01$ thousand dollars, are:

Period 1: $500(2{,}260/3{,}400)(0.600)=199.41$ thousand dollars.
Period 2: $1{,}500(2{,}260/3{,}400)(0.200)=199.41$ thousand dollars.
Period 3: $2{,}000(2{,}260/3{,}400)(0)=0.00$ thousand dollars.

For comparison, the individual ratios of losses to adjusted exposure are $120/200=60.00\%$, $840/1{,}200=70.00\%$, and $1{,}300/2{,}000=65.00\%$. Their unweighted average is $65.00\%$, not the Cape Cod result. The pooled ratio gives greater weight to periods with more adjusted exposure. The mistake avoided is averaging period ratios equally when the method calls for pooling losses and adjusted exposures.

Assumptions, Checks, and Exam Traps

Cape Cod’s common ratio is useful when the accident periods are sufficiently comparable for a shared expected loss ratio to be meaningful. In practice, the analyst must consider whether the exposure measures and loss amounts reflect compatible coverage, limits, deductibles, and valuation bases. The method also relies on suitable emergence proportions: if those proportions are wrong, the adjusted exposures and period-level unreported estimates will be affected.

The experience is pooled, but each period retains its own exposure and maturity. This is why a single common ratio does not imply identical reserves across periods. A period with higher exposure or a larger unreported proportion will generally receive a larger estimated unreported amount, all else equal. Also keep the estimated unreported amount distinct from an estimated ultimate: on the same cumulative-loss basis, the latter is the observed cumulative loss plus the estimated unreported amount.

Using the wrong denominator. The Cape Cod denominator is $\sum_i E_i p_i$, not $\sum_i E_i$, when $p_i$ represents the proportion emerged by the valuation date.
Mixing paid and incurred bases. Paid losses require paid emergence proportions; incurred losses require incurred emergence proportions. Pairing one basis with the other distorts the adjustment.
Applying the wrong proportion. The common ratio is applied to $E_i(1-p_i)$ for unreported losses, not to $E_i p_i$.
Replacing the observed losses. The estimated unreported amount is added to observed cumulative losses to form an estimated ultimate; it does not replace those observed losses.

Key takeaway: Cape Cod estimates one experience-based loss ratio from pooled cumulative losses and maturity-adjusted exposure. It then estimates each period’s remaining losses as exposure times that common ratio times the period’s unreported proportion.

Check Your Understanding

For each problem, use a consistent loss basis and report monetary answers in thousands of dollars to two decimal places where needed.

A hypothetical account has two accident periods. Their exposures are $500$ and $1{,}000$, emergence proportions are $0.600$ and $0.800$, and cumulative losses are $180$ and $640$. Find the Cape Cod ratio. (Answer: $74.55\%$)
Two periods have exposures $800$ and $1{,}000$, emergence proportions $0.500$ and $0.750$, and cumulative losses $200$ and $600$. Find the common ratio and estimated unreported loss for each period. (Answer: ratio $69.57\%$; unreported losses $278.26$ and $173.91$)
A hypothetical portfolio has exposures $400$, $1{,}200$, and $2{,}400$; emergence proportions $0.500$, $0.750$, and $1.000$; and cumulative losses $120$, $630$, and $1{,}320$. Find the pooled ratio and the total estimated unreported loss. (Answer: ratio $59.14\%$; total unreported loss $295.71$)
On a paid basis, two periods have exposures $600$ and $1{,}000$, paid emergence proportions $0.250$ and $0.750$, and cumulative paid losses $90$ and $525$. Find the Cape Cod ratio and total estimated unreported paid losses. (Answer: ratio $68.33\%$; total unreported paid losses $478.33$)

What You'll Learn