Aggregating Expected Losses Across Classes and Coverages

Why Expected Losses Must Be Aggregated Carefully

An experience rating plan may cover several classifications and several coverages. To establish its expected-loss benchmark, the actuary needs the expected loss for each applicable combination and then a total for the account or plan. The arithmetic is usually straightforward; the difficult part is confirming that every input refers to a compatible exposure, time period, and definition of loss.

For example, a rate per payroll unit cannot be multiplied by vehicle-years, and a loss estimate on a gross basis should not be added to an estimate on a net basis without reconciling them. Aggregation is reliable only after these bases have been checked.

Definition: A classification-coverage cell is one combination of a risk classification and a coverage. Let $E_c$ denote the exposure for classification $c$, $r_{c,k}$ the expected-loss rate for coverage $k$ in that classification, and $L_{c,k}$ the resulting expected loss. When the rate is expressed per unit of that exposure and the inputs share the same loss and time basis, $$ L_{c,k}=E_c r_{c,k}. $$ The total expected loss across the included classifications and coverages is $$ L_{\mathrm{total}}=\sum_c\sum_k L_{c,k}. $$

The expression $E_c r_{c,k}$ assumes that the rate and exposure match: for instance, a rate per $1,000 of payroll must be multiplied by payroll measured in $1,000 units. It also assumes that the rate already represents the intended expected-loss definition and period. If an input rate is for a different valuation date, coverage scope, or loss basis, it must be reconciled before using it in the sum.

What Must Be Consistent Before Summing

A practical calculation starts with a cell-by-cell inventory, not a grand total. For every expected-loss estimate, identify the classification, coverage, exposure amount and unit, period, and loss basis. For example, distinguish losses that are limited from unlimited losses, or losses net of recoveries from losses before recoveries. These are different expected-loss measures; adding them as if they were alike can distort the benchmark.

The exposure unit does not have to be the same for every classification if each cell is calculated correctly. Payroll units for one class and vehicle-years for another can both produce expected losses in dollars. The resulting dollar amounts can be added when their coverage, period, valuation, and loss definitions are aligned. What cannot generally be done is to pool the unlike exposure units and report one meaningful rate per “unit.”

A combined rate is useful only when its denominator is meaningful for every included amount. If all classes use the same exposure unit, the total expected loss divided by total exposure gives the exposure-weighted rate. It is not generally the simple average of the class rates. If the classes use unlike units, keep the expected-loss amounts separate by unit rather than creating a misleading pooled rate.

Why Expected Losses Add

Let $X_{c,k}$ be the random loss for classification $c$ and coverage $k$, measured on the same intended basis. The account loss is the sum of the included cell losses. The expectation of that sum is the sum of the cell expectations. This result does not require the losses to be independent.

E\left[\sum_c\sum_k X_{c,k}\right] =\sum_c\sum_k E[X_{c,k}]

This follows from the additivity of expectation: for two losses, $E[X+Y]=E[X]+E[Y]$, and applying that identity repeatedly gives the finite sum. Dependence between coverages can affect the variance of total loss, but it does not change this expected-value identity. The practical qualification is that the components must truly partition or otherwise correctly represent the total being measured; overlapping estimates can count the same loss twice.

Worked Examples

Worked Example: Combining Classes on a Common Exposure Basis

A hypothetical program has one coverage and three classifications. All rates are expected losses per payroll unit of $1,000, for the same annual period and valuation basis. Class A has 1,200 payroll units and a rate of $18.00; Class B has 800 units and a rate of $31.00; Class C has 500 units and a rate of $24.00. Calculate total expected loss and the combined rate per payroll unit.

Method and assumptions. Multiply each class exposure by its own rate, then sum the cell estimates. The exposures and rates are already aligned to the same annual period and coverage basis; the rates are treated as given expected-loss estimates. No particular claim distribution is required for this calculation, and no interest discounting or mortality assumption is involved.

Class A: $1{,}200 \times \$18.00=\$21{,}600.00$.

Class B: $800 \times \$31.00=\$24{,}800.00$.

Class C: $500 \times \$24.00=\$12{,}000.00$.

Total expected loss: $\$21{,}600.00+\$24{,}800.00+\$12{,}000.00=\$58{,}400.00$. Total exposure is $1{,}200+800+500=2{,}500$ payroll units.

Combined rate: $\$58{,}400.00/2{,}500=\$23.36$ per payroll unit. As a check, $2{,}500\times \$23.36=\$58{,}400.00$, matching the sum of the class estimates.

The method avoids the common mistake of taking an unweighted average of the three rates. That average is $(\$18+\$31+\$24)/3=\$24.33$ per unit, which would imply $2{,}500\times \$24.33=\$60{,}825.00$ using the rounded rate. It gives each class equal influence despite their different exposures, so it does not reproduce the total expected loss.

Worked Example: Aggregating a Two-Coverage Rate Table

A hypothetical account has two classifications, and both coverages use the same payroll-unit exposure. Class A has 900 units; its expected-loss rates are $12.40 for Coverage 1 and $5.60 for Coverage 2. Class B has 600 units; its rates are $20.10 and $8.30, respectively. All rates are per payroll unit and refer to the same annual valuation and loss basis. Find expected loss by coverage and in total.

Method and assumptions. Calculate each classification-coverage cell independently, then add first within coverage and then across coverages. The rates are given expected-loss estimates; there are no separate adjustments for trend, development, or limits in this calculation.

Class A, Coverage 1: $900\times \$12.40=\$11{,}160.00$. Class A, Coverage 2: $900\times \$5.60=\$5{,}040.00$. Class A contributes $\$11{,}160.00+\$5{,}040.00=\$16{,}200.00$.

Class B, Coverage 1: $600\times \$20.10=\$12{,}060.00$. Class B, Coverage 2: $600\times \$8.30=\$4{,}980.00$. Class B contributes $\$12{,}060.00+\$4{,}980.00=\$17{,}040.00$.

Coverage 1 total: $\$11{,}160.00+\$12{,}060.00=\$23{,}220.00$. Coverage 2 total: $\$5{,}040.00+\$4{,}980.00=\$10{,}020.00$. Therefore, total expected loss is $\$23{,}220.00+\$10{,}020.00=\$33{,}240.00$.

Total exposure is $900+600=1{,}500$ payroll units. The combined rate is $\$33{,}240.00/1{,}500=\$22.16$ per payroll unit. The coverage-specific combined rates are $\$23{,}220.00/1{,}500=\$15.48$ and $\$10{,}020.00/1{,}500=\$6.68$; their sum, $\$22.16$, checks against the total rate. The calculation treats the two coverage components as additive and non-overlapping for this purpose.

Worked Example: Adding Expected Losses With Unlike Exposure Units

A hypothetical insurance account includes a vehicle classification measured in vehicle-years and a contracting classification measured in payroll units of $1,000. For the same annual period and liability coverage basis, the vehicle class has 48 vehicle-years at an expected-loss rate of $245 per vehicle-year. The contracting class has 2,500 payroll units at a rate of $13.80 per payroll unit. Calculate the total expected loss and decide whether a single combined rate per exposure unit can be reported.

Method and assumptions. Calculate the expected loss within each class using its own matched exposure unit, then add the dollar amounts. The two rates are given expected-loss estimates for the same liability coverage, annual period, valuation, and loss basis. No frequency or severity distribution is needed to multiply exposure by rate.

Vehicle class: $48\times \$245.00=\$11{,}760.00$.

Contracting class: $2{,}500\times \$13.80=\$34{,}500.00$.

Total expected loss: $\$11{,}760.00+\$34{,}500.00=\$46{,}260.00$. This is the sum of expected losses in dollars, so adding is valid after confirming the shared period and loss basis.

A single combined rate per exposure unit is not defined by the data: vehicle-years and payroll units are unlike denominators. Dividing $\$46{,}260.00$ by $48+2{,}500$ would treat the two unlike units as interchangeable and yield a rate with no coherent exposure meaning. The check is to retain the separate products, $\$11{,}760.00$ and $\$34{,}500.00$, and confirm that their sum equals the reported total.

Common Errors and a Practical Check

The most frequent aggregation errors are not failures of multiplication; they are failures to preserve the meaning of the inputs. A rate table can look uniform while hiding different exposure units, periods, coverage scopes, or loss bases. Before adding, verify those dimensions and inspect whether each class-coverage cell has been included once and only once.

Using an unweighted rate average. Rates from classes with different exposure volumes need exposure weighting if a combined rate is required and the denominator is common.
Combining unlike denominators. Add expected-loss dollars when their loss bases agree, but do not manufacture a pooled rate from incompatible exposure units.
Mixing loss definitions. Confirm that all estimates use the intended treatment of limits, deductibles, recoveries, and expenses before aggregation.
Double counting a component. Check that coverage estimates are defined as additive components of the target total, rather than overlapping estimates of the same losses.
Summing before checking period alignment. Rates and exposures must describe the intended period and valuation basis. Reconcile mismatched inputs before calculating the aggregate.

A useful final check is to calculate totals in two directions when the table permits it: sum cells by classification, then across classifications; separately sum by coverage, then across coverages. Both routes should produce the same total. If they do not, a cell may be missing, duplicated, or incorrectly multiplied.

Check Your Understanding

For each problem, preserve the stated exposure and loss bases. Dollar answers are rounded to cents where applicable.

A hypothetical class has 700 payroll units at $11.50 per unit, and a second has 300 payroll units at $18.50 per unit. Find total expected loss and the combined rate per payroll unit. (Answer: $13,600.00 total; $13.60 per unit.)
For Coverage 1, Class A has 400 common exposure units at $7.25 per unit and Class B has 600 at $9.50. For Coverage 2, the respective rates are $2.75 and $4.00. All rates share the same period and loss basis. Find expected loss by coverage and in total. (Answer: Coverage 1, $8,600.00; Coverage 2, $3,500.00; total, $12,100.00.)
A vehicle class has 30 vehicle-years at $240 per vehicle-year. A separate class has 18,000 payroll units at $0.80 per payroll unit. The estimates refer to the same coverage and period. Find total expected loss and state whether one combined rate per exposure unit is meaningful. (Answer: $21,600.00 total; no, because vehicle-years and payroll units are unlike denominators.)
Two classes share one exposure unit. Class A has 250 units, with rates of $16 for Coverage 1 and $4 for Coverage 2. Class B has 350 units, with rates of $9 and $6. Calculate total expected loss and the proportion of it attributable to Coverage 1. (Answer: $10,250.00 total; Coverage 1 is $7,150.00, or 69.76%.)
A hypothetical account has 1,200 common exposure units in Class A and 800 in Class B. Coverage 1 rates are $5.20 and $8.10; Coverage 2 rates are $2.30 and $3.40. The rates are per unit and share the same annual loss basis. Calculate each coverage total and the account total. (Answer: Coverage 1, $12,720.00; Coverage 2, $5,480.00; account total, $18,200.00.)

What You'll Learn