Segment-Level Expected Loss Estimates

Why Estimate Losses by Segment?

An expected ultimate loss estimate is an a priori estimate of the total losses expected to emerge for a specified group of claims, before relying on that group’s own fully developed experience. The group may be a coverage, class, territory, or other defined segment. Estimating separately matters when segments have different expected loss levels: a single assumption applied indiscriminately can obscure those differences and give an inappropriate estimate for an individual segment.

This tutorial focuses on calculating a separate estimate for each defined segment. It does not estimate unpaid losses by subtracting emerged losses, and it does not combine the separate estimates into an overall total. Those are distinct calculations. The estimates here may use either a premium-based expected loss ratio or an exposure-based expected loss cost, following the earlier material on expected ultimate losses and matching assumptions to their bases.

Definition: A segment-level expected ultimate loss estimate is the expected ultimate loss for a specified segment, calculated using an expected loss assumption and the matching premium or exposure measure for that segment. For segment $j$, write the estimate as $E[L_j]$, where $L_j$ denotes that segment’s ultimate losses.

A segment must have a clear scope. For example, “Coverage A in Territory 1” is more precise than “preferred business” if that label does not identify which policies it includes. For any given calculation, use the same definition of segment for the premium or exposure measure and for its expected loss assumption. If a premium amount includes one coverage but its expected loss ratio applies to another, the multiplication may be arithmetically correct but actuarially meaningless.

Two Segment-Level Calculation Bases

On a premium basis, multiply the segment’s premium by its expected loss ratio. The premium and expected loss ratio must be stated on compatible bases: for example, earned premium matched with an expected loss ratio intended for earned premium. Any stated scope differences, such as whether losses include loss adjustment expenses (LAE), must also be treated consistently.

E[L_j] = P_j \times ELR_j

Here, $P_j$ is the premium for segment $j$, and $ELR_j$ is the expected loss ratio selected for that same segment and premium basis. On an exposure basis, multiply the segment’s exposure count or amount by its expected loss cost per corresponding exposure unit:

E[L_j] = X_j \times ELcost_j

Here, $X_j$ denotes segment $j$’s exposures, and $ELcost_j$ is the expected loss cost per exposure unit for that segment. If the exposure unit is a policy-year, for instance, the loss cost must also be expressed per policy-year. The estimate is in dollars when the exposure count is multiplied by a dollar loss cost per exposure.

Assumptions and limits: These formulas use expected losses, not observed losses, as the basis for the estimate. The selected expected loss ratio or loss cost must apply to the same segment, valuation scope, and measurement basis as the premium or exposures. The calculation itself does not require a claim-frequency or claim-severity distribution, a mortality assumption, or discounting. If a task specifies discounting or a particular loss scope, incorporate it only as directed and keep the assumption consistent across the inputs.

Why the Premium Formula Holds

The premium-based formula follows directly from the definition of an expected loss ratio. For segment $j$, suppose the selected expected loss ratio is defined as the ratio of the expected ultimate loss to the matching premium:

ELR_j = \frac{E[L_j]}{P_j}, \qquad P_j > 0.

Multiplying both sides by $P_j$ gives $P_j ELR_j = E[L_j]$. Thus, the formula is not a separate prediction rule added to the definition; it rearranges the definition of the expected loss ratio. It is appropriate only when that ratio refers to the same premium measure and segment used in the calculation. A ratio based on one segment or premium basis cannot be substituted merely because it is available.

Define the segment.
State exactly which class, coverage, or other grouping the estimate covers.

Select the matching basis.
Identify the premium or exposure measure for that segment, including its units and scope.

Apply that segment’s assumption.
Multiply premium by the segment’s expected loss ratio, or exposures by its expected loss cost per exposure unit.

Check the result.
Confirm that the units produce dollars and that no other segment’s premium, exposures, or assumption entered the calculation.

Worked Calculations

Each example treats the selected expected loss assumptions as already established. No interest rate or mortality assumption is specified or used, no discounting is applied, and no probability distribution is needed: these are direct calculations of expected losses from the stated assumptions.

Worked Example: Premium-Based Estimates for Two Coverages

A hypothetical equipment insurer estimates ultimate losses separately for two coverages. Coverage A has $1,250,000 of earned premium and a selected expected loss ratio of 58%. Coverage B has $840,000 of earned premium and a selected expected loss ratio of 72%. Both ratios apply to earned premium and include the same loss scope. Calculate each coverage’s expected ultimate loss.

Method. Use the premium-based formula separately for each coverage. Because the premium and ratio are both coverage-specific and on the same basis, no adjustment or conversion is needed.

Coverage A. Convert 58% to 0.58 and multiply:

E[L_A] = \$1{,}250{,}000 \times 0.58 = \$725{,}000.

Coverage B. Convert 72% to 0.72 and multiply:

E[L_B] = \$840{,}000 \times 0.72 = \$604{,}800.

Check. Each estimate is less than its corresponding premium because each expected loss ratio is less than 100%. The resulting estimates are $725,000 for Coverage A and $604,800 for Coverage B. No interest, mortality, or distribution parameter is used.

Common mistake avoided. Applying Coverage A’s 58% ratio to Coverage B’s premium would produce $487,200, but it would use an assumption selected for the wrong segment. The calculation must pair each coverage’s premium with that coverage’s expected loss ratio.

Worked Example: Exposure-Based Estimates for Two Classes

A hypothetical property program distinguishes two building classes. Class 1 has 2,400 insured building-years and an expected loss cost of $185 per building-year. Class 2 has 1,100 insured building-years and an expected loss cost of $310 per building-year. Calculate the expected ultimate loss for each class.

Method. Use exposure count multiplied by expected loss cost per exposure unit. The stated loss costs are both per building-year, matching the exposure units.

Class 1.

E[L_1] = 2{,}400\ \text{building-years} \times \$185\ \text{per building-year} = \$444{,}000.

Class 2.

E[L_2] = 1{,}100\ \text{building-years} \times \$310\ \text{per building-year} = \$341{,}000.

Check. The units cancel to dollars in each calculation. The estimates are $444,000 for Class 1 and $341,000 for Class 2. No interest, mortality, or loss distribution is assumed or required.

Common mistake avoided. Using $185 for both classes would give Class 2 an estimate of $203,500, but the problem assigns Class 2 its own expected loss cost of $310 per building-year. Matching the exposure count to the correct segment-specific loss cost is essential.

Worked Example: Selecting a Separate Basis for Each Segment

A hypothetical insurer has two separately defined segments. Segment R has $960,000 of earned premium and an expected loss ratio of 64% selected for that segment. Segment S has 1,750 earned policy-years and an expected loss cost of $420 per policy-year. The inputs have compatible scope within each segment. Calculate each segment’s expected ultimate loss using its stated basis.

Method. Segment R has premium and an expected loss ratio, so use the premium formula. Segment S has exposures and an expected loss cost, so use the exposure formula. Do not apply one segment’s basis or assumption to the other.

Segment R. Convert 64% to 0.64:

E[L_R] = \$960{,}000 \times 0.64 = \$614{,}400.

Segment S.

E[L_S] = 1{,}750\ \text{policy-years} \times \$420\ \text{per policy-year} = \$735{,}000.

Check. The premium-based calculation yields dollars, and the policy-year units cancel in the exposure-based calculation, also yielding dollars. The separate expected ultimate loss estimates are $614,400 for Segment R and $735,000 for Segment S.

Common mistake avoided. A common error is to treat the two formulas as interchangeable even when the stated inputs differ. For example, Segment S has no expected loss ratio in the problem, so multiplying its exposures by a percentage would not be justified. Use the measure and assumption actually supplied for each segment.

Basis Checks and Exam Traps

Before accepting a result, check more than the multiplication. The arithmetic is usually straightforward; the main risk is pairing inputs that do not belong together. A segment-level estimate should answer a specific question: “What is the expected ultimate loss for this segment, given this segment’s premium or exposures and this segment’s selected expectation?”

Using a pooled assumption for a distinct segment. A ratio or loss cost selected for a different class or coverage may not represent the segment being estimated. Use the assumption assigned to the segment.
Mixing units. A loss cost per vehicle-year cannot be multiplied directly by vehicle-months without converting the exposure basis. Check that the numerator and denominator units align.
Mixing premium bases. An expected loss ratio defined for earned premium should not be applied to a premium amount on another basis unless the problem supplies the appropriate conversion or confirms comparability.
Ignoring scope. A loss ratio that includes LAE and a target estimate defined to exclude LAE do not have matching loss scope. Use consistent definitions.
Confusing expected ultimate losses with unpaid losses. The formulas estimate ultimate losses. They do not subtract paid or reported losses. An estimate of unpaid losses requires a separate calculation.
Rounding too early. Retain the stated precision in the ratio or loss cost during the multiplication, then round the final dollar estimate as requested.

A quick reasonableness check can catch errors. Under a premium-based method, an expected loss ratio of 50% implies an expected loss estimate equal to half the matching premium. Under an exposure-based method, doubling the exposure count while holding the same segment loss cost fixed doubles the estimate. These checks do not prove that the selected assumption is appropriate, but they can reveal misplaced decimal points, wrong units, or accidental use of another segment’s input.

Key takeaway: Calculate each defined segment on its own matching basis: $E[L_j] = P_j \times ELR_j$ for a premium-based estimate, or $E[L_j] = X_j \times ELcost_j$ for an exposure-based estimate. Verify segment identity, units, and scope before interpreting the result.

Check Your Understanding

For each problem, calculate the requested segment-level expected ultimate loss. Do not combine separate segment estimates.

A hypothetical marine coverage has $1,360,000 of earned premium and a matching expected loss ratio of 62.5%. What is its expected ultimate loss? (Answer: $850,000)
A hypothetical warranty class has 3,250 contract-years and an expected loss cost of $148 per contract-year. What is its expected ultimate loss? (Answer: $481,000)
Two hypothetical liability segments have the following earned premiums and expected loss ratios: Segment X, $720,000 and 47%; Segment Y, $460,000 and 69%. Calculate the separate expected ultimate loss for each. (Answer: Segment X, $338,400; Segment Y, $317,400)
A hypothetical property portfolio has two building classes. Class A has 840 building-years and an expected loss cost of $525 per building-year. Class B has 1,200 building-years and an expected loss cost of $390 per building-year. Calculate each class’s estimate. (Answer: Class A, $441,000; Class B, $468,000)
A hypothetical insurer estimates Segment M using $875,000 of earned premium and a 56% expected loss ratio. It estimates Segment N using 1,600 earned policy-years and a $275 expected loss cost per policy-year. Calculate both estimates using the appropriate basis for each segment. (Answer: Segment M, $490,000; Segment N, $440,000)

What You'll Learn