Calculating Cape Cod Ultimates and Reserves

Introduction

The Cape Cod method estimates a common loss ratio from the experience that has emerged so far, while allowing for the fact that accident periods are at different stages of development. A period with only a small share of its losses reported contributes less reported exposure to the estimate than a mature period. Once the ratio is calculated, it is used with each period’s estimated unreported share to calculate unpaid losses.

The calculation has several distinct steps: establish a consistent exposure and loss basis, convert development information into a reported proportion, calculate the Cape Cod loss ratio, and then calculate expected losses, unpaid losses, and ultimate losses by period. Keeping these quantities separate prevents a common error: treating the expected ultimate amount as though it were automatically the final estimate of that period’s ultimate.

Definitions and Calculation Framework

For accident period $i$, let $E_i$ denote the exposure measure used in the calculation, $C_i$ the losses reported to date, and $p_i$ the estimated proportion of ultimate losses reported. The exposure basis and reported losses must be consistent across periods. For example, if $E_i$ is premium, the resulting loss ratio is on that premium basis.

Definition: The Cape Cod loss ratio, denoted here by $\hat r_{\mathrm{CC}}$, is total reported losses divided by total exposure adjusted for the proportion reported: $$ \hat r_{\mathrm{CC}}=\frac{\sum_i C_i}{\sum_i E_i p_i}. $$ The expected ultimate loss amount for period $i$ is $E_i\hat r_{\mathrm{CC}}$. The estimated unpaid amount is $E_i\hat r_{\mathrm{CC}}(1-p_i)$, and the estimated ultimate is reported losses plus estimated unpaid losses: $$ \widehat{\mathrm{Unpaid}}_i=E_i\hat r_{\mathrm{CC}}(1-p_i), \qquad \hat U_i=C_i+\widehat{\mathrm{Unpaid}}_i. $$

If the available development information is a factor to ultimate $f_i$, and it is defined as ultimate losses divided by reported losses, then the reported proportion is $p_i=1/f_i$. The unreported proportion is $1-p_i$, not $f_i-1$. For a fully reported period, $p_i=1$, so its estimated unpaid amount is zero under this calculation.

The framework assumes that the loss amounts, exposure measures, and development proportions refer to compatible bases and accident periods. It also uses one common Cape Cod loss ratio across the periods included. If different periods have different exposure bases or the development information does not represent the losses being analyzed, the calculation is not valid without first resolving that mismatch.

Why the Loss-Ratio Formula Uses Adjusted Exposure

The denominator $\sum_i E_i p_i$ represents the exposure corresponding to the reported share of losses. To see why, suppose that a common loss ratio $r$ applies across periods. The expected ultimate loss for period $i$ is then $E_i r$, and the expected reported amount is $E_i r p_i$. Summing those expected reported amounts gives $r\sum_i E_i p_i$. Equating this to the observed total reported losses and solving for $r$ gives the Cape Cod ratio:

\sum_i C_i \approx r\sum_i E_i p_i \quad\Longrightarrow\quad \hat r_{\mathrm{CC}}=\frac{\sum_i C_i}{\sum_i E_i p_i}.

This is a calibration: the common ratio is chosen so the ratio-based expected reported losses, summed across periods, equal the observed reported losses. It does not require each period’s reported amount to equal its own ratio-based expected reported amount.

From Expected Losses to Unpaid and Ultimate Losses

The expected ultimate amount $E_i\hat r_{\mathrm{CC}}$ is a ratio-based benchmark. The unpaid estimate is the part of that benchmark associated with the estimated unreported share. The final period ultimate estimate then adds that unpaid amount to the period’s actual reported losses. These steps are related but are not interchangeable.

Find the reported proportion.
Use the supplied proportion, or convert a factor to ultimate using $p_i=1/f_i$.

Calculate the Cape Cod ratio.
Divide total reported losses by $\sum_i E_i p_i$.

Calculate ratio-based expected ultimate losses.
For each period, calculate $E_i\hat r_{\mathrm{CC}}$.

Calculate unpaid and final ultimate losses.
Multiply expected ultimate losses by $1-p_i$ for unpaid losses, then add reported losses to obtain the final ultimate estimate.

Worked Example: Calculating a Ratio and Reserve for Two Periods

A hypothetical specialty insurer has exposure amounts of $100$ and $200$ for two accident periods. The estimated reported proportions are $0.40$ and $0.80$, and reported losses are $30$ and $100$, respectively. Calculate the Cape Cod loss ratio, expected ultimate losses, unpaid losses, and final ultimate losses for each period. Amounts are in thousands of dollars.

Method. Calculate the reported-exposure denominator first, then calculate the common ratio and apply the unreported proportions.

The denominator is

$$ 100(0.40)+200(0.80)=40+160=200. $$

Total reported losses are $30+100=130$. Therefore,

\hat r_{\mathrm{CC}}=\frac{130}{200}=0.65.

For period 1, ratio-based expected ultimate losses are $100(0.65)=65$. The unreported proportion is $1-0.40=0.60$, so unpaid losses are $65(0.60)=39$. The final ultimate estimate is $30+39=69$.

For period 2, ratio-based expected ultimate losses are $200(0.65)=130$. The unreported proportion is $1-0.80=0.20$, so unpaid losses are $130(0.20)=26$. The final ultimate estimate is $100+26=126$.

Check. Total unpaid losses are $39+26=65$, and total final ultimate losses are $69+126=195$. Independently, total reported losses plus total unpaid losses are $130+65=195$. The aggregate ratio-based expected ultimate is also $65+130=195$, as the calibration implies.

Common mistake avoided: The expected ultimate for period 1 is $65$, but its final ultimate estimate is $69$, because the method adds the actual reported amount to the unpaid estimate. The expected ultimate is not substituted for the reported-plus-unpaid result period by period.

Working from Development Factors

A factor to ultimate must be converted before calculating the ratio. If a factor is $2.00$, the estimated reported proportion is $1/2.00=0.50$; if the factor is $1.25$, the reported proportion is $1/1.25=0.80$. Applying a factor directly as though it were a proportion can make the denominator too large and can produce an impossible unreported proportion.

Worked Example: Converting Factors and Calculating Total Unpaid Losses

A hypothetical portfolio has exposure amounts of $150$ and $250$, reported losses of $40$ and $120$, and factors to ultimate of $2.00$ and $1.25$. Calculate the reported proportions, Cape Cod loss ratio, unpaid losses by period, and total ultimate losses. Amounts are in thousands of dollars. Round final amounts and the ratio to three decimal places.

Method. Convert each factor to a reported proportion, then use the same adjusted-exposure calculation for the ratio.

The reported proportions are

p_1=\frac{1}{2.00}=0.50,\qquad p_2=\frac{1}{1.25}=0.80.

The adjusted exposure denominator is $150(0.50)+250(0.80)=75+200=275$. Total reported losses are $40+120=160$. Thus

\hat r_{\mathrm{CC}}=\frac{160}{275}=0.581818\ldots

For period 1, the unreported proportion is $1-0.50=0.50$. Its unpaid estimate is

150(0.581818\ldots)(0.50)=43.636\ldots

For period 2, the unreported proportion is $1-0.80=0.20$. Its unpaid estimate is

250(0.581818\ldots)(0.20)=29.091\ldots

Total unpaid losses are $43.636\ldots+29.091\ldots=72.727\ldots$. Total ultimate losses are reported losses plus unpaid losses: $160+72.727\ldots=232.727\ldots$. Rounded as requested, the ratio is $0.582$, period unpaid amounts are $43.636$ and $29.091$, total unpaid losses are $72.727$, and total ultimate losses are $232.727$, all amounts in thousands.

Check. The two periods’ ratio-based expected ultimate amounts are $150(0.581818\ldots)=87.273\ldots$ and $250(0.581818\ldots)=145.455\ldots$. Their sum, $232.727\ldots$, equals the total final ultimate. The unpaid total is also $0.581818\ldots[150(0.50)+250(0.20)]=72.727\ldots$.

Common mistake avoided: Using $150+250=400$ as the denominator ignores that only a portion of each period’s losses is reported. The correct denominator is $275$, not $400$.

Calculating Period Results and Reconciling Totals

The Cape Cod ratio is estimated from all included periods together, but each period’s unpaid amount depends on its own exposure and reported proportion. A fully reported period can contribute reported losses and adjusted exposure to the ratio while having no additional unpaid amount. Its reported losses are still included when calculating the final ultimate total.

Worked Example: Three Accident Periods with Different Maturities

A hypothetical program has exposure amounts of $100$, $200$, and $300$. Its reported proportions are $0.50$, $0.75$, and $1.00$, and reported losses are $25$, $90$, and $180$. Calculate the loss ratio, period expected ultimate losses, unpaid losses, and final ultimate losses. Amounts are in thousands of dollars. Round final amounts and the ratio to three decimal places.

Method. Include the fully reported third period in the ratio calculation, but assign it zero unpaid losses because its unreported proportion is zero.

The adjusted exposure is $100(0.50)+200(0.75)+300(1.00)=50+150+300=500$. Total reported losses are $25+90+180=295$. Therefore,

\hat r_{\mathrm{CC}}=\frac{295}{500}=0.590.

The ratio-based expected ultimate amounts are:

100(0.590)=59,\qquad 200(0.590)=118,\qquad 300(0.590)=177.

The unreported proportions are $0.50$, $0.25$, and $0$. Hence unpaid amounts are

59(0.50)=29.500,\qquad 118(0.25)=29.500,\qquad 177(0)=0.

Add each unpaid amount to its corresponding reported losses. The final ultimate estimates are $25+29.500=54.500$, $90+29.500=119.500$, and $180+0=180.000$. Total unpaid losses are $59.000$, and total ultimate losses are $295+59=354.000$, all in thousands.

Check. Total ratio-based expected ultimate losses are $59+118+177=354$, equal to the sum of the final accident-period ultimate estimates, $54.500+119.500+180.000=354.000$. The third period’s final estimate is its reported amount because it is fully reported.

Common mistake avoided: The first period’s final ultimate is $54.500$, not its ratio-based expected ultimate of $59$. Its actual reported amount is $25$, and its estimated unpaid amount is $29.500$; the final estimate is their sum.

Common Errors and Practical Checks

The aggregate reconciliation is a useful calculation check. From the definition of the ratio,

\sum_i E_i\hat r_{\mathrm{CC}}p_i =\hat r_{\mathrm{CC}}\sum_i E_i p_i =\sum_i C_i.

Therefore, summing the ratio-based expected ultimate amounts gives

\sum_i E_i\hat r_{\mathrm{CC}} =\sum_i E_i\hat r_{\mathrm{CC}}p_i+ \sum_i E_i\hat r_{\mathrm{CC}}(1-p_i) =\sum_i C_i+\sum_i\widehat{\mathrm{Unpaid}}_i.

This is why aggregate expected ultimate losses equal aggregate reported losses plus aggregate unpaid losses. The equality is an aggregate reconciliation; it does not say that each period’s ratio-based expected ultimate equals that period’s final reported-plus-unpaid estimate.

Using unadjusted exposure. The denominator is $\sum_i E_i p_i$, not $\sum_i E_i$. Omitting $p_i$ disregards differences in maturity.
Reversing the factor conversion. With a factor to ultimate $f_i$, use $p_i=1/f_i$, then $1-p_i$ for the unreported proportion.
Leaving out a mature period. A fully reported period has $p_i=1$; it contributes its exposure and reported losses to the ratio calculation, although its unpaid estimate is zero.
Double counting reported losses. The unpaid estimate is added to reported losses once to obtain the final ultimate. Do not add reported losses to the ratio-based expected ultimate amount.
Rounding too early. Carry the ratio at full precision through the period calculations and round only the final displayed results.

Key takeaway: Calculate the Cape Cod ratio using total reported losses divided by exposure weighted by the reported proportion. Then calculate each period’s unpaid amount from its own unreported share, and obtain its final ultimate by adding unpaid losses to reported losses.

Check Your Understanding

For each problem, keep the reported proportion, expected ultimate, unpaid estimate, and final ultimate distinct. Amounts are in thousands of dollars unless stated otherwise.

A hypothetical portfolio has exposures $90$ and $150$, reported proportions $0.50$ and $0.75$, and reported losses $27$ and $90$. Find the Cape Cod ratio and total unpaid losses. (Answer: ratio $0.7429$; total unpaid $61.2857$.)
Two accident periods have exposures $100$ and $200$, reported losses $35$ and $140$, and factors to ultimate $2.00$ and $1.25$. Find the reported proportions, ratio, total reserve, and total ultimate. (Answer: proportions $0.50, 0.80$; ratio $0.8333$; reserve $75.0000$; ultimate $250.0000$.)
Exposures are $50,100,150$, reported proportions are $0.40,0.60,1.00$, and reported losses are $10,45,130$. Find the ratio and unpaid estimate for each period. (Answer: ratio $0.8043$; unpaid amounts $24.1304,32.1739,0.0000$.)
For the data in question 3, calculate the final ultimate estimate by period and the aggregate ultimate. (Answer: period ultimates $34.1304,77.1739,130.0000$; aggregate ultimate $241.3043$.)
A two-period portfolio has exposures $80$ and $120$, reported proportions $0.50$ and $1.00$, and reported losses $20$ and $100$. A student divides total reported losses by total exposure. Find the correct ratio and total unpaid amount, and identify the student’s denominator error. (Answer: correct ratio $0.7500$; unpaid $30.0000$; use adjusted exposure $40+120=160$, not total exposure $200$.)

What You'll Learn