Expected Claims as an A Priori Ultimate Loss Method

What Makes an Expected Loss Estimate A Priori?

An ultimate loss estimate answers how much loss is expected for a defined group of policies or exposures over the full period of coverage, including losses that have not yet been reported or paid. An a priori expected claims method estimates that total from a selected expected loss ratio or expected loss cost and a volume measure. It does not use the group’s emerged claim amounts as an input.

The method is useful when claim experience is unavailable, immature, or intentionally excluded from the estimate. “Independent of emerged losses” describes the calculation: the estimate is determined by the assumed expected loss measure and the relevant premium or exposure volume. It does not mean the estimate is certain, or that the selected assumption is independent of all information. The expected loss ratio or cost may reflect prior analysis and judgment.

Definition: An a priori expected ultimate loss estimate is an estimate of ultimate losses for a specified group, based on an expected loss ratio applied to a premium base or an expected loss cost applied to an exposure base, without using the group’s emerged losses in the calculation.

Two Volume Bases

Before applying a formula, specify the group being estimated and the loss basis. The expected loss measure and the volume measure must refer to compatible coverage, time periods, limits, deductibles, and loss components. For example, if the expected loss ratio includes allocated loss adjustment expenses, the resulting estimate includes them; if the ratio is for indemnity losses only, the estimate should not be described as including expenses.

For the premium-based approach, let $EP$ be earned premium for the group and let $r$ be its selected a priori expected loss ratio, expressed as a decimal. The model assumes that the group’s expected ultimate losses are $r$ dollars per dollar of earned premium. Under that assumption:

\widehat{L}^{\mathrm{ult}}_{\mathrm{AP}}=r\,EP

For the exposure-based approach, let $N$ be the number of exposure units and let $c$ be the selected expected ultimate loss cost per exposure unit. The model assumes the same expected loss cost is applicable to each counted unit, or that $c$ is an appropriate average for the mix of units. Then:

\widehat{L}^{\mathrm{ult}}_{\mathrm{AP}}=c\,N

The quantity being estimated is the same in both formulas: ultimate losses for the specified group. The distinction is the volume measure used to scale the expected loss. The premium method is natural when expected losses are expressed relative to premium; the exposure method is natural when a meaningful exposure unit and expected cost per unit are available.

Worked Example: Premium-Based A Priori Estimate

A hypothetical commercial property portfolio has earned premium of $1,800,000 for a defined accident period. The selected expected loss ratio is 0.62. At the valuation date, $487,000 of losses have emerged. Estimate ultimate losses using the a priori expected loss ratio method.

Method. Apply the supplied expected loss ratio to earned premium. The $487,000 of emerged losses is not an input to this a priori estimate.

Step 1: Confirm the units. The expected loss ratio is 0.62 dollars of ultimate loss per dollar of earned premium. The earned premium is $1,800,000.

Step 2: Calculate the estimate.

\widehat{L}^{\mathrm{ult}}_{\mathrm{AP}} =0.62(1{,}800{,}000) =1{,}116{,}000

The a priori ultimate loss estimate is $1,116,000. The calculation does not add the emerged losses to this amount: those losses are part of the ultimate total being estimated, not an additional component of it.

Check and trap avoided. The result is 62% of earned premium, since $1{,}116{,}000/1{,}800{,}000=0.62$. A common error is to use written premium if it differs from earned premium. For example, applying 0.62 to hypothetical written premium of $2,000,000 would give $1,240,000, but that is not the requested calculation when the stated basis is earned premium.

Why the Formulas Work

These formulas follow from how the expected loss measure is defined. For the premium method, suppose the selected model assigns expected ultimate losses equal to $r$ times the earned premium for each part of the group. If the group consists of parts indexed by $j$, with earned premium $EP_j$, then its expected loss under the model is $rEP_j$. Adding the expected losses across the parts gives:

\sum_j rEP_j=r\sum_j EP_j=r\,EP

For the exposure method, suppose each exposure unit has expected ultimate loss $c$. The expected loss for $N$ units is the sum of the expected losses for the units, so:

\sum_{j=1}^{N} c=cN

The key condition is not that every policy will produce exactly the selected ratio or cost. Individual losses vary. Rather, the parameter is being used as the expected loss per unit of the chosen volume base for the group. If that parameter is not appropriate for the group, multiplying it precisely by the volume still produces an inappropriate estimate.

Worked Example: Exposure-Based A Priori Estimate

A hypothetical equipment protection program covers 8,400 machine-years. A machine-year is one insured machine covered for one year. The selected expected ultimate loss cost is $73.25 per machine-year. By the valuation date, $201,400 of losses have emerged. Estimate ultimate losses using the exposure-based method.

Method. Multiply the number of covered machine-years by the expected ultimate loss cost per machine-year.

Step 1: Check that the units match. The cost is $73.25 per machine-year, and the volume is 8,400 machine-years.

Step 2: Multiply the cost by the exposure count.

\widehat{L}^{\mathrm{ult}}_{\mathrm{AP}} =73.25(8{,}400) =615{,}300

The a priori ultimate loss estimate is $615,300. The emerged amount of $201,400 does not enter this calculation.

Check and trap avoided. Dividing the estimate by 8,400 gives $73.25 per machine-year, confirming that the units and arithmetic are consistent. A common mistake is to multiply by 8,400 policies if the data actually give 8,400 machine-years. The exposure count must be measured in the same unit used to define the expected loss cost.

Estimating Several Segments

A single expected loss ratio or expected loss cost may not be suitable for a group with materially different components. If each segment has its own appropriate parameter, calculate the a priori estimate by segment and add the estimates. The parameters should be applied to their matching bases; a ratio selected for one segment should not be applied indiscriminately to another.

For a premium-based estimate across $k$ segments, where segment $j$ has earned premium $EP_j$ and expected loss ratio $r_j$, the combined estimate is:

\widehat{L}^{\mathrm{ult}}_{\mathrm{AP}} =\sum_{j=1}^{k} r_j EP_j

This is equivalent to applying a premium-weighted average ratio to total earned premium. It is not generally equivalent to taking an unweighted arithmetic average of the segment ratios.

Worked Example: Combining Segments With Different Ratios

A hypothetical insurer estimates ultimate losses for two separately classified groups. Group A has earned premium of $900,000 and an a priori expected loss ratio of 0.48. Group B has earned premium of $600,000 and an a priori expected loss ratio of 0.72. Estimate combined ultimate losses.

Method. Apply each expected loss ratio only to its corresponding earned premium, then add the segment estimates.

Step 1: Estimate losses for Group A.

$$ 0.48(900{,}000)=432{,}000 $$

Step 2: Estimate losses for Group B.

$$ 0.72(600{,}000)=432{,}000 $$

Step 3: Add the estimates.

$$ 432{,}000+432{,}000=864{,}000 $$

Combined a priori ultimate losses are $864,000. Total earned premium is $1,500,000, so the implied combined expected loss ratio is $864{,}000/1{,}500{,}000=0.576$.

Check and trap avoided. Applying the implied ratio gives $0.576(1{,}500{,}000)=864{,}000$, a reconciliation check. An unweighted average of the two ratios is 0.60, which would produce $900,000. That is the wrong weighting because the segments have different earned premium volumes.

Assumptions, Interpretation, and Common Traps

An a priori estimate is only as relevant as its parameter and base. The calculation assumes that the selected expected loss ratio or loss cost applies to the group being estimated, and that the premium or exposure volume represents the same coverage and period. For a loss ratio, earned premium must be on the basis intended by the selected ratio. For a loss cost, the exposure unit must be defined consistently with the data used to set that cost.

Define the target.
Specify the policy group, accident or coverage period, and whether losses include any specified loss adjustment expenses.

Choose the matching base.
Use earned premium with an expected loss ratio, or a clearly defined exposure count with an expected loss cost.

Apply and aggregate.
Multiply each base by its corresponding parameter, then add segment estimates when the group has multiple components.

Label the result correctly.
The result is an estimate of ultimate losses, not automatically an estimate of unpaid losses or a claim liability at the valuation date.

The last distinction matters. Ultimate losses refer to the full loss total for the defined group. An unpaid loss estimate concerns the portion not yet paid. If ultimate losses and paid losses are measured on the same basis, unpaid losses can be estimated as ultimate losses less paid losses. That subtraction is a separate step and uses paid loss information; it does not change the fact that the a priori ultimate estimate itself was calculated without emerged losses.

Using the wrong premium measure. Do not substitute written premium for earned premium unless the stated method and parameter specifically call for written premium.
Using an inconsistent exposure unit. A per-vehicle-year loss cost requires vehicle-years, not a count of vehicles that ignores coverage duration.
Adding emerged losses to the a priori ultimate estimate. The estimate already represents the expected total ultimate losses; adding observed losses double counts them conceptually.
Averaging segment ratios without weights. When segment premium differs, the combined ratio must reflect those differences.
Treating the estimate as a prediction of exact losses. It is a model-based expected amount. Actual ultimate losses may be above or below it.

Key takeaway: An a priori expected claims method scales a selected expected loss measure by a compatible volume base. The premium version uses expected loss ratio times earned premium; the exposure version uses expected loss cost times exposure units. Neither calculation uses the group’s emerged loss amount.

Check Your Understanding

For each problem, identify the applicable base and calculate or interpret the a priori estimate.

A hypothetical policy group has earned premium of $740,000 and a selected expected loss ratio of 0.55. What is its a priori expected ultimate loss estimate? (Answer: $407,000.)
A hypothetical service contract portfolio has 2,500 covered contract-years and an expected ultimate loss cost of $41.20 per contract-year. Calculate the a priori ultimate loss estimate. (Answer: $103,000.)
Two hypothetical segments have earned premiums of $400,000 and $1,100,000, with expected loss ratios of 0.65 and 0.40, respectively. Calculate the combined a priori ultimate losses. (Answer: $700,000.)
A hypothetical portfolio has 3,200 exposure units and an expected loss cost of $125 per unit. An analyst instead multiplies by 3,200 policies, although each policy covers two exposure units. What is the correct estimate, assuming the exposure count is therefore 6,400 units? (Answer: $800,000.)
A hypothetical group has earned premium of $2,400,000 and an expected loss ratio of 0.58. It has $760,000 of emerged losses at valuation. What is the a priori ultimate estimate, and should the emerged amount be added to it? (Answer: $1,392,000; no, the emerged amount is not added.)

What You'll Learn