Expected Losses as the Plan Benchmark

Why a Plan Needs an Expected-Loss Benchmark

An experience-rating plan compares a risk’s observed losses with a baseline for the losses expected from its exposure. The benchmark gives meaning to the observation: actual losses of $50,000 may be unusually high for a small exposure but quite ordinary for a much larger one. The comparison is useful only when the expected and actual amounts refer to the same risk, coverage, exposure basis, and experience period.

In this tutorial, expected losses are the benchmark amount for covered losses over the selected experience period. They are not the risk’s actual losses, and they are not the full premium. A premium can include expenses and other components in addition to the expected cost of covered losses. The benchmark is used to interpret experience; by itself, it does not specify how a plan converts that comparison into a premium change.

Definition: Let $L$ be the aggregate covered loss amount for a specified risk and experience period. The expected losses for that risk and period are $E[L]$, calculated on a stated coverage and exposure basis. If the expected loss cost is $r$ dollars per exposure unit and the matching exposure is $e$ units, then the benchmark is $e r$. For multiple segments with different applicable expected loss costs, the benchmark is the sum of the segment amounts.

The benchmark is an expectation, not a prediction that the risk will incur exactly that amount. Individual claims are uncertain, so actual losses can be above or below expected losses even when the assumptions are appropriate. The comparison describes how observed losses relate to the chosen baseline; it does not establish that the observed result will continue.

Calculating the Benchmark and Comparing Experience

When an expected loss cost is stated per a scaled exposure unit, first convert the observed exposure to that unit. For example, a rate of $32 per $1,000 of payroll applies to payroll measured in thousands of dollars. If payroll is $1,250,000, the matching exposure is $1{,}250{,}000/1{,}000=1{,}250$ units, not 1,250,000 units.

Two common summaries compare actual losses $A$ with expected losses $E$. The actual-to-expected ratio is $A/E$, provided $E>0$. A ratio of 1.00 means actual losses equal expected losses; a ratio above 1.00 means actual losses exceed the benchmark. The dollar deviation is $A-E$, which is positive when actual losses exceed expected losses and negative when they fall below it. These are descriptive comparisons, not by themselves plan adjustments.

\text{Expected losses}=e r,\qquad \text{actual-to-expected ratio}=\frac{A}{E},\qquad \text{dollar deviation}=A-E

Worked Example: Converting Payroll to the Rate’s Exposure Unit

A hypothetical employer has $1,250,000 of payroll in a specified experience period. The expected loss cost for the employer’s classification and coverage is $32 per $1,000 of payroll. Suppose this rate reflects an expected frequency of 0.02 claims per $1,000 of payroll and an average covered claim severity of $1,600. Actual covered losses for the period are $48,600. Find expected losses, the actual-to-expected ratio, and the dollar deviation.

Method and assumptions. Use the supplied expected loss cost and convert payroll into units of $1,000. As a simple model underlying the rate, let claim count per $1,000 unit be Poisson with mean 0.02 and let each claim have fixed severity $1,600. The expected loss cost per unit is therefore $0.02(1{,}600)=\$32$. Assume the rate and actual losses refer to the same period, coverage, and loss valuation basis. Interest is 0%; no discounting is applied. Mortality is not applicable to this property-and-casualty loss example.

Step 1: Convert payroll. The number of exposure units is $1{,}250{,}000/1{,}000=1{,}250$. The implied expected claim count is $1{,}250(0.02)=25$.

Step 2: Calculate expected losses. By exposure times expected loss cost, $E=1{,}250(\$32)=\$40{,}000$. Equivalently, the expected claim count times fixed severity is $25(\$1{,}600)=\$40{,}000$.

Step 3: Compare actual with expected. The ratio is $\$48{,}600/\$40{,}000=1.215$. The dollar deviation is $\$48{,}600-\$40{,}000=+\$8{,}600$.

Check and trap avoided. The ratio implies actual losses are 21.5% above the benchmark, and $0.215(\$40{,}000)=\$8{,}600$, agreeing with the dollar deviation. A common unit error is to use 1,250,000 directly as the number of $1,000 units, which overstates expected losses by a factor of 1,000.

Combining Segments and Periods

A risk’s experience may contain multiple classifications, exposure types, or time segments. If each segment has a suitable expected loss cost, calculate its benchmark using that segment’s exposure and then add the amounts. This avoids applying one rate to exposure that belongs to a different classification or period.

E_{\mathrm{total}}=\sum_{j=1}^{n} e_j r_j

This addition follows from the linearity of expectation. If total covered loss is the sum of segment losses, $L=\sum_j L_j$, then $E[L]=E[\sum_j L_j]=\sum_j E[L_j]$. Independence between segment losses is not required for this equality. The calculation still requires each segment’s expected loss cost to apply to its own exposure and to the coverage and period represented by its loss amount.

Worked Example: Combining Two Classifications

A hypothetical contractor has 80 unit-years in classification A and 120 unit-years in classification B during the same experience period. For the same specified coverage, expected frequency and fixed average severity are 0.10 claims per unit-year and $4,500 per claim in A, and 0.05 claims per unit-year and $6,000 per claim in B. Actual covered losses across both classifications are $81,000. Calculate the total expected losses and compare actual with expected.

Method and assumptions. Calculate each segment’s expected loss cost as frequency times average severity, multiply by its own exposure, and add the benchmarks. For each classification, assume Poisson claim counts with the stated mean frequency per unit-year and fixed claim severity at the stated amount. No interest discounting is used (interest rate 0%); mortality is not applicable. Assume the two classifications have the same coverage, experience period, and loss valuation basis as the actual losses. Independence is not needed to add the expectations.

Step 1: Classification A. Its expected loss cost is $0.10(\$4{,}500)=\$450$ per unit-year. Expected losses are $80(\$450)=\$36{,}000$. The implied expected claim count is $80(0.10)=8$, and $8(\$4{,}500)=\$36{,}000$.

Step 2: Classification B. Its expected loss cost is $0.05(\$6{,}000)=\$300$ per unit-year. Expected losses are $120(\$300)=\$36{,}000$. The implied expected claim count is $120(0.05)=6$, and $6(\$6{,}000)=\$36{,}000$.

Step 3: Add and compare. Total expected losses are $\$36{,}000+\$36{,}000=\$72{,}000$. The actual-to-expected ratio is $\$81{,}000/\$72{,}000=1.125$, and the dollar deviation is $\$81{,}000-\$72{,}000=+\$9{,}000$.

Check and trap avoided. A ratio of 1.125 is 12.5% above expected, and $0.125(\$72{,}000)=\$9{,}000$. Applying classification A’s $450 cost to all 200 unit-years would give $90,000, but that incorrectly treats the classification B exposure as if it belonged to classification A.

What Makes the Comparison Valid?

An arithmetic result is not a valid benchmark comparison unless the expected losses and actual losses refer to matching quantities. The following assumptions should be checked before interpreting a ratio or deviation.

Match the coverage.
The expected loss cost and actual losses must represent the same covered causes of loss, limits, deductibles, and other material coverage terms. If one amount includes losses excluded from the other basis, the comparison is not like-for-like.

Match the exposure basis.
Use the exposure quantity and unit for which the expected loss cost is defined. Confirm scale factors, such as dollars of payroll per $1,000, and assign each exposure segment to the applicable classification or rate.

Match the period.
Expected losses must cover the same experience period as the actual losses. If exposure or expected loss costs differ by period, calculate each segment separately before adding.

Match the loss basis.
Use an actual-loss measure consistent with the expected-loss benchmark, including its stated treatment of valuation and loss adjustments. Do not compare amounts prepared on different bases without first reconciling them.

A key justification for segmentation is that expectations are period- and exposure-specific. Suppose a risk has exposure $e_1$ in one period at expected loss cost $r_1$ and exposure $e_2$ in another at cost $r_2$. The expected aggregate loss is the expectation of the sum of the period losses. By linearity, it is $e_1r_1+e_2r_2$. It is generally not correct to multiply total exposure by one of the rates unless that rate is appropriate for all of the exposure, or to use an appropriately weighted average rate that reproduces the same sum.

Worked Example: Preserving Period-Specific Benchmarks

A hypothetical manufacturer’s experience period consists of three annual segments, all with the same $1 million per-occurrence coverage limit and the same exposure basis of $1,000 of payroll. Payroll is $2,000,000, $2,500,000, and $3,000,000 in years 1, 2, and 3, respectively. The corresponding expected loss costs are $4.00, $4.20, and $4.50 per exposure unit. Actual covered losses for all three years, valued on the same stated basis, total $38,400. Calculate expected losses and the comparison.

Method and assumptions. Convert each year’s payroll to $1,000 units, apply that year’s cost, and sum. For an illustrative frequency-severity interpretation, take the claim count in each unit as Poisson with mean 0.002, 0.0021, and 0.00225 in years 1, 2, and 3, respectively, with fixed severity $2,000. These give loss costs of $4.00, $4.20, and $4.50. Interest is 0%, so amounts are not discounted; mortality is not applicable. The actual losses and benchmark are assumed to use identical coverage, periods, and valuation treatment.

Step 1: Convert exposures. The annual exposures are $2{,}000{,}000/1{,}000=2{,}000$, $2{,}500{,}000/1{,}000=2{,}500$, and $3{,}000{,}000/1{,}000=3{,}000$ units.

Step 2: Calculate each period’s expected losses. Year 1: $2{,}000(\$4.00)=\$8{,}000$. Year 2: $2{,}500(\$4.20)=\$10{,}500$. Year 3: $3{,}000(\$4.50)=\$13{,}500$.

Step 3: Sum and compare. Expected losses total $\$8{,}000+\$10{,}500+\$13{,}500=\$32{,}000$. The ratio is $\$38{,}400/\$32{,}000=1.200$, and the dollar deviation is $\$38{,}400-\$32{,}000=+\$6{,}400$.

Check and trap avoided. The deviation is 20.0% of $32,000, so it agrees with the ratio of 1.200. Using the first year’s $4.00 rate for all 7,500 exposure units would produce $30,000, not the correct $32,000, because the expected loss costs vary by year. Also, if the actual losses were measured under a different limit, the ratio would not be interpretable as a like-for-like comparison without reconciling the coverage basis.

Common Mistakes and Practical Interpretation

The most damaging errors often occur before the ratio is calculated. A misplaced exposure-unit conversion can change the benchmark by a factor of 10, 100, or 1,000. A coverage mismatch can make a mathematically correct ratio answer the wrong question. An experience-period mismatch can likewise compare actual losses from one set of dates with expected losses from another.

Using the wrong exposure scale. If the rate is per $1,000 of payroll, divide payroll by 1,000 before multiplying by the rate.
Applying one segment’s cost to all exposure. When expected costs vary by classification or period, calculate each segment separately and add.
Confusing expected losses with premium. Expected losses represent the expected covered-loss component on the stated basis, not necessarily the total amount charged.
Interpreting the ratio as a plan adjustment. The ratio describes actual relative to expected. Any rule translating it into a rating outcome is a separate plan feature.
Ignoring coverage differences. Comparing losses under one limit or deductible with expected losses under another can distort the benchmark even when every arithmetic step is correct.

Key takeaway: Expected losses are a meaningful experience benchmark only when exposure, coverage, experience period, and loss basis align. Calculate segment-specific expected amounts as exposure times the applicable expected loss cost, add them, and then compare actual losses with that total.

Check Your Understanding

For each problem, calculate the expected-loss benchmark first. Then use the requested comparison measure, checking units and assumptions.

A hypothetical repair portfolio has $875,000 of payroll and an expected loss cost of $18 per $1,000 of payroll. Actual covered losses are $18,900. Find expected losses, the actual-to-expected ratio, and the dollar deviation. (Answer: Expected losses $15,750; ratio 1.2000; deviation +$3,150.)
A hypothetical fleet has 55 vehicle-years, an expected frequency of 0.062 claims per vehicle-year, and fixed average severity of $5,000. Actual covered losses are $15,500. Find expected losses, the ratio, and the deviation. Assume the actual losses use the same coverage and period; no interest discounting or mortality assumption applies. (Answer: Expected losses $17,050; ratio 0.9091; deviation −$1,550.)
A hypothetical account has 40 unit-years in class A, with frequency 0.08 and fixed severity $2,500, and 70 unit-years in class B, with frequency 0.04 and fixed severity $4,000. Actual losses are $21,120 on the same coverage and period. Find the total benchmark and comparison. (Answer: Expected losses $19,200; ratio 1.1000; deviation +$1,920.)
For a hypothetical risk, payroll is $1,200,000 in year 1 and $800,000 in year 2. Expected loss costs are $5.50 and $6.00 per $1,000 of payroll, respectively. Actual covered losses for both years total $12,540. Find expected losses and the ratio and deviation. (Answer: Expected losses $11,400; ratio 1.1000; deviation +$1,140.)
A hypothetical account has 240 exposure units, and its expected loss cost is $240 per 1,000 exposure units. Actual losses are $63,360 on the matching basis. Find the benchmark and comparison. (Answer: Expected losses $57.60; ratio 1,100; deviation +$63,302.40.)

What You'll Learn