Expected Losses Across Multiple Experience Years

Why Aggregate Expected Losses Across Years?

Experience rating often uses losses from more than one experience year. The exposure volume may differ from year to year, and the expected loss rate may differ as well. A valid aggregate benchmark must account for both. Adding annual expected losses does this directly; averaging annual rates without considering exposure generally does not.

The calculation is simple when each year has a comparable exposure measure and an applicable expected loss rate. The important judgment is whether the analysis needs only the total benchmark or also needs a benchmark for each individual year. An overall expected loss total cannot, by itself, preserve the differences among years.

Definition: For experience year $y$, let $e_y$ be the exposure on the selected basis and $r_y$ the expected loss rate per exposure unit, both applicable to that year. The year's expected losses are $E_y=e_y r_y$. Across $n$ years, total exposure is $\sum_{y=1}^{n} e_y$ and aggregate expected losses are $\sum_{y=1}^{n} E_y$. If all years use the same comparable exposure basis, the aggregate expected loss rate is $\bar r=\frac{\sum_{y=1}^{n}e_y r_y}{\sum_{y=1}^{n}e_y}$.

These calculations are on an expected-value basis. They do not require an interest rate, mortality assumption, or a specified probability distribution for individual losses. The rates and exposures are assumed to be appropriate for the same loss definition and experience-year basis. If the exposure units or the loss basis are not comparable, expected dollars may still be summed when each year's amount has been calculated consistently, but a single aggregate rate may not be meaningful.

Deriving the Aggregate Rate

The aggregate rate is an exposure-weighted average, not usually the ordinary arithmetic average of the annual rates. The reason follows directly from the definition of expected losses: each year's rate contributes in proportion to the exposure to which it applies.

Calculate each year's expected losses.
For year $y$, multiply the year's exposure by its applicable rate: $E_y=e_y r_y$.

Add the annual expected losses.
The total expected losses are $E_{\mathrm{total}}=\sum_{y=1}^{n} e_y r_y$.

Divide by total exposure when a pooled rate is needed.
Since total expected losses equal total exposure times the rate on the combined basis, $\bar r=E_{\mathrm{total}}/\sum_{y=1}^{n} e_y$.

\bar r = \frac{\sum_{y=1}^{n} e_y r_y}{\sum_{y=1}^{n} e_y} \qquad\text{and}\qquad E_{\mathrm{total}} = \bar r\sum_{y=1}^{n}e_y = \sum_{y=1}^{n}E_y

To justify the formula, suppose a single rate $\bar r$ is to reproduce exactly the expected losses from all years combined. By definition, it must satisfy $\bar r\sum_y e_y=\sum_y E_y$. Substituting $E_y=e_y r_y$ and dividing by positive total exposure gives the weighted-average formula. Thus, the pooled rate preserves the total expected dollars, but it does not necessarily preserve each year's expected dollars.

Worked Examples

Worked Example: Calculating a Pooled Rate from Unequal Exposures

A hypothetical commercial repair program has three completed experience years. Each exposure unit represents one stated unit of insured activity, and all rates are in dollars per exposure unit. Year 1 has 1,000 exposure units at a rate of $42; Year 2 has 1,500 units at $48; and Year 3 has 800 units at $45. Find the expected losses by year, total expected losses, and the aggregate rate.

Method and assumptions. Apply each rate to its matching year's exposure, then sum. The exposure basis and expected-loss definition are consistent across the years. This is an expected-loss calculation; no interest rate, mortality assumption, or individual-loss distribution is needed.

Year 1: $1{,}000\times \$42=\$42{,}000$.

Year 2: $1{,}500\times \$48=\$72{,}000$.

Year 3: $800\times \$45=\$36{,}000$.

Total exposure is $1{,}000+1{,}500+800=3{,}300$ units. Total expected losses are $\$42{,}000+\$72{,}000+\$36{,}000=\$150{,}000$. Therefore, the aggregate rate is $\$150{,}000/3{,}300=\$45.4545\ldots$ per unit, or $45.45 per unit rounded to the nearest cent.

Check. Multiplying the unrounded rate by total exposure gives $(\$150{,}000/3{,}300)\times3{,}300=\$150{,}000$, as required. The rate is between the lowest annual rate, $42, and the highest, $48.

Common mistake avoided. The ordinary average of the rates is $(42+48+45)/3=\$45$. Multiplying that by 3,300 gives only $148,500, not $150,000. That average gives each year's rate equal influence despite the different exposure volumes.

Worked Example: Comparing Annual and Aggregate Loss Ratios

For a hypothetical equipment service portfolio, the expected loss inputs are 2,400 exposure units at $310 per unit in Year 1, 1,800 units at $340 in Year 2, and 3,000 units at $325 in Year 3. Actual losses are $700,000, $650,000, and $900,000, respectively. Calculate each year's expected losses and loss ratio, then calculate the aggregate expected losses and aggregate loss ratio. Define each loss ratio as actual losses divided by expected losses.

Method and assumptions. Compute each expected amount using its year's own rate. Compare actual to expected at the year level before combining. All figures use the same loss basis and exposure units; no interest, mortality, or individual-loss distribution assumption enters the calculation.

Year 1: Expected losses are $2{,}400\times\$310=\$744{,}000$. The loss ratio is $\$700{,}000/\$744{,}000=0.940860\ldots$, or 0.9409 to four decimals.

Year 2: Expected losses are $1{,}800\times\$340=\$612{,}000$. The loss ratio is $\$650{,}000/\$612{,}000=1.062091\ldots$, or 1.0621.

Year 3: Expected losses are $3{,}000\times\$325=\$975{,}000$. The loss ratio is $\$900{,}000/\$975{,}000=0.923076\ldots$, or 0.9231.

The total expected losses are $\$744{,}000+\$612{,}000+\$975{,}000=\$2{,}331{,}000$. Total actual losses are $\$700{,}000+\$650{,}000+\$900{,}000=\$2{,}250{,}000$. The aggregate loss ratio is $\$2{,}250{,}000/\$2{,}331{,}000=0.96525\ldots$, or 0.9653 to four decimals.

Check. The aggregate ratio is below 1 because total actual losses of $2,250,000 are below total expected losses of $2,331,000. The annual ratios differ, but their combined result is determined by the summed dollars, not by an unweighted average of the three ratios.

Common mistake avoided. Averaging the three rounded annual ratios equally would not generally give the aggregate ratio. The correct aggregate calculation weights each year's ratio by its expected losses: the total actual losses divided by the total expected losses.

Worked Example: Why Year-Specific Benchmarks May Be Needed

A hypothetical account has two experience years on the same exposure basis. Year 1 has 2,000 units at an expected loss rate of $45 per unit, and actual losses of $99,000. Year 2 has 1,000 units at $60 per unit, and actual losses of $54,000. Calculate the expected losses and actual-to-expected ratios by year, then show what changes if the pooled rate is used to assign expected losses back to each year.

Method and assumptions. First calculate the annual benchmarks at their applicable rates. Then derive a pooled rate that preserves the combined expected amount. This comparison is useful when a later calculation needs year-level results. No interest rate, mortality assumption, or individual-loss distribution is required.

Year 1 benchmark: $2{,}000\times\$45=\$90{,}000$. Its actual-to-expected ratio is $\$99{,}000/\$90{,}000=1.10$.

Year 2 benchmark: $1{,}000\times\$60=\$60{,}000$. Its actual-to-expected ratio is $\$54{,}000/\$60{,}000=0.90$.

Total exposure is 3,000 units and total expected losses are $\$90{,}000+\$60{,}000=\$150{,}000$. The pooled rate is $\$150{,}000/3{,}000=\$50$ per unit. If that rate is used to reassign expected losses by year, Year 1 becomes $2{,}000\times\$50=\$100{,}000$, and Year 2 becomes $1{,}000\times\$50=\$50{,}000$. The reassigned amounts still total $150,000, but differ from the original year-specific benchmarks by +$10,000 and −$10,000.

Check. The overall actual-to-expected ratio remains $(\$99{,}000+\$54{,}000)/\$150{,}000=1.02$ under either allocation because the pooled expected total is unchanged. However, the year-specific ratios become $0.99$ and $1.08$ using the pooled rate, rather than 1.10 and 0.90. Pooling has changed the year-level comparisons.

Common mistake avoided. A pooled rate is suitable for a combined benchmark when the exposure bases are comparable, but it should not replace annual rates if the task requires each year’s benchmark or year-specific comparison.

When to Retain the Annual Rates

The aggregate expected loss total is appropriate when the question concerns the combined experience-period benchmark. Retain the year-specific expected amounts when the calculation or decision depends on identifying which year produced which benchmark. Examples include reporting annual comparisons, applying a year-specific adjustment, or reconciling annual inputs to an aggregate total.

A pooled rate is valid only when its denominator is a meaningful total exposure on a common basis. If one year's rate is per vehicle-year and another year's rate is per payroll unit, adding the exposures and dividing total expected losses by that sum does not produce a meaningful single rate. Calculate each year's expected dollars using its own rate and exposure basis, then sum the expected dollars if the loss definitions are compatible.

The same caution applies when different rates reflect different rating bases or expected-loss definitions. The arithmetic of a total remains straightforward, but a pooled rate should be reported only when its units and interpretation are clear. Aggregation preserves total dollars; it does not make unlike inputs interchangeable.

Common Errors and Checks

Taking a simple average of rates. Use an exposure-weighted rate when exposures differ. An ordinary average is appropriate only if the exposure weights are equal or if the task explicitly asks for an unweighted average.
Applying one rate to all years without checking applicability. A pooled rate can reproduce a total and still misstate individual-year benchmarks.
Mixing exposure units. Confirm that the denominator of a proposed aggregate rate has a consistent unit.
Adding rates instead of expected dollars. Calculate $e_y r_y$ for each year, then sum those expected amounts.

A useful final check is that total expected losses equal both the sum of annual expected losses and, when the exposure basis is common, the pooled rate multiplied by total exposure. Also check that the pooled rate lies between the smallest and largest annual rates when all exposures are positive. If either check fails, revisit the arithmetic, units, or inputs.

Key takeaway: Calculate expected losses year by year, then add the dollar amounts. Use an exposure-weighted rate only when the years share a comparable exposure basis, and retain annual rates whenever year-level benchmarks matter.

Check Your Understanding

For each problem, calculate annual expected losses before combining them. Dollar amounts are hypothetical, and all rates are stated on the applicable exposure basis.

A program has 400 exposure units at $28 per unit in Year 1 and 650 units at $32 in Year 2. Find total expected losses and the aggregate rate per unit, rounded to the nearest cent. (Answer: $32,000 total; $30.48 per unit.)
A portfolio has 900 units at $75 per unit, 1,200 units at $70, and 700 units at $82.50. Find total exposure, total expected losses, and the aggregate rate to the nearest cent. (Answer: 2,800 units; $209,250; $74.73 per unit.)
Year 1 has 500 exposure units at $120 and actual losses of $66,000. Year 2 has 750 units at $100 and actual losses of $72,000. Find each year's expected losses and actual-to-expected ratio, and the aggregate ratio to four decimals. (Answer: Expected losses $60,000 and $75,000; annual ratios 1.1000 and 0.9600; aggregate ratio 1.0222.)
Two years have 2,000 and 1,000 exposure units, at rates of $45 and $60 per unit. Find the pooled rate and the expected losses assigned to each year if that pooled rate is applied to both. How do those amounts compare with the original year-specific expectations? (Answer: Pooled rate $50; pooled allocations $100,000 and $50,000, versus original expectations $90,000 and $60,000.)
Year 1 has 1,200 units at $50 per unit. Year 2 has an unknown exposure at $65 per unit. If total expected losses are $190,000, find the missing exposure and the aggregate rate, rounded to the nearest cent. (Answer: 2,000 units; $59.38 per unit.)

What You'll Learn