Why the Basis Matters
An a priori expected-claims estimate can be expressed using either premium or exposure as its volume measure. The two approaches can target the same expected ultimate losses, but they require different inputs: one needs a loss ratio and a premium amount; the other needs a loss cost per exposure unit and an exposure count. Choosing the wrong denominator or mixing incompatible units can produce a plausible-looking but incorrect estimate.
This distinction is useful when selecting an expected-claims method for immature experience. The estimate must be stated for a defined coverage, accident period, and loss measure. The premium or exposure used as its base must describe the same risk volume and be consistent with the expected loss assumption. The methods below are aligned with the CAS Exam 5 (Basic Ratemaking and Estimated Claim Liabilities) syllabus.
The word expected is important: \(r\) and \(c\) are prospective assumptions for the losses being estimated, not necessarily ratios or averages calculated from the current immature losses. The premium method scales a loss expectation by dollars of premium. The exposure method scales it by units such as vehicle-years, policy-years, payroll units, or another clearly specified measure of insured risk.
Inputs, Units, and Matching
For either approach, first define the target loss amount. It might be losses for a specified coverage and accident period, for example, before or after a stated loss adjustment, deductible, or limit treatment. Then select the volume basis and expected value that correspond to that same target. A loss ratio for one coverage should not be applied to premium for a different coverage; a loss cost per vehicle-year should not be multiplied by a count of policies unless each policy represents exactly one vehicle-year.
| Feature | Premium-based approach | Exposure-based approach |
|---|---|---|
| Expected value | Expected loss ratio, \(r\) | Expected loss cost, \(c\) |
| Volume input | Relevant premium, \(P\), in dollars | Relevant exposure, \(E\), in exposure units |
| Units of expected value | Loss dollars per premium dollar | Loss dollars per exposure unit |
| Expected loss calculation | \(rP\) | \(cE\) |
| Typical unit check | Dimensionless ratio × premium dollars = loss dollars | Dollars per unit × exposure units = loss dollars |
The premium amount must be explicitly identified. For an accident-period estimate, earned premium is often the natural volume measure because it corresponds to coverage provided during that period. If a problem instead specifies another premium measure, use it as stated and verify that the expected ratio is defined on that basis. Premiums need not all be identical for the calculation, but the total premium must represent the same target population and coverage as the losses.
Similarly, exposure is not just a number of contracts. It is a count or measure with a defined unit and time basis. A vehicle insured for six months may represent 0.5 vehicle-years under a time-proportional convention; it is not automatically one full vehicle-year. Expected loss costs must also correspond to the chosen exposure unit and the target loss definition.
These relationships explain why the two methods can agree when their assumptions describe the same risks. If \(P\) is the matching premium and \(E\) the matching exposure, then a consistent pair of assumptions satisfies \(rP=cE\). This is a consistency condition, not a claim that any loss ratio and loss cost can be interchanged without information about the premium and exposure relationship.
How the Two Calculations Work
For a single homogeneous group, the calculations are direct. For multiple groups, calculate each group using its own corresponding base and then add the expected losses. A total expected loss ratio should not be formed by taking an unweighted average of group ratios; its natural aggregate is based on total expected losses divided by total premium. The same principle applies to exposure-based loss costs: totals depend on exposure weights.
Specify the coverage, period, and loss measure whose ultimate amount is being estimated.
Use a premium-based assumption with a premium amount, or an exposure-based assumption with exposure measured in the matching units.
Multiply the expected loss ratio by premium, or multiply expected loss cost by exposure. Confirm that the result is in loss dollars.
For distinct groups, calculate their expected losses separately and sum those dollar amounts.
The unit check is also a short justification for the calculation. A loss ratio is expected loss dollars divided by premium dollars, so multiplying it by the matching premium dollars recovers expected loss dollars. A loss cost is expected loss dollars per exposure unit, so multiplying it by the number of those units recovers expected loss dollars. Additivity then gives the group total: if \(L_j\) is the expected loss for group \(j\), the combined expectation is \(\sum_j L_j\).
Worked Example: Premium-Based Estimate for Two Coverage Groups
A hypothetical insurer is estimating expected ultimate losses for two coverage groups in a defined accident period. Group A has $480,000 of earned premium and an expected loss ratio of 0.62. Group B has $320,000 of earned premium and an expected loss ratio of 0.55. The assumptions are stated to apply to the same coverage and loss definition as the premiums. Calculate expected losses by group and in total.
Method. Apply each group’s premium-based expected loss ratio to that group’s premium. Since a loss ratio is dimensionless, each product is in dollars.
Group A: \(0.62 \times \$480{,}000=\$297{,}600\).
Group B: \(0.55 \times \$320{,}000=\$176{,}000\).
Total: \(\$297{,}600+\$176{,}000=\$473{,}600\). The combined premium is \(\$480{,}000+\$320{,}000=\$800{,}000\), so the implied aggregate expected loss ratio is \(\$473{,}600/\$800{,}000=0.592\), or 59.2%.
Check and mistake avoided. The implied ratio reproduces the total: \(0.592 \times \$800{,}000=\$473{,}600\). A simple average of 62% and 55% would be 58.5%, which is wrong here because the groups have different premium volumes. The premium-weighted aggregate is the appropriate check.
Worked Example: Exposure-Based Estimate With Partial-Year Units
A hypothetical small-fleet program has an expected loss cost of $145 per vehicle-year for a specified coverage and loss measure. Its exposure consists of 240 vehicles insured for a full year and 80 vehicles insured for six months. Using a time-proportional vehicle-year convention, calculate the total exposure and expected ultimate losses.
Method. Convert every exposure to vehicle-years before applying the loss cost. A full year contributes 1 vehicle-year; six months contributes \(6/12=0.5\) vehicle-year.
Full-year exposure: \(240 \times 1=240\) vehicle-years.
Six-month exposure: \(80 \times 0.5=40\) vehicle-years.
Total exposure: \(240+40=280\) vehicle-years.
Expected ultimate losses: \(\$145\text{ per vehicle-year}\times280\text{ vehicle-years}=\$40{,}600\).
Check and mistake avoided. The units cancel to dollars. Counting all 320 vehicles as full-year exposures would give \(320 \times \$145=\$46{,}400\), overstating the estimate by \(\$5{,}800\). That calculation incorrectly treats 80 half-year exposures as 80 vehicle-years instead of 40.
Worked Example: Selecting the Usable Basis From Mixed Inputs
A hypothetical book has 1,200 appliance-policy-years and 900 service-contract policy-years. For the defined target loss measure, the expected loss costs are $38 per appliance-policy-year and $24 per service-contract policy-year. The corresponding earned premiums are $84,000 and $45,000. Calculate expected losses using the exposure basis, and then determine whether one combined expected loss ratio can be obtained from the information given without adding an assumption.
Method. Use each expected loss cost only with its matching exposure count. The expected loss cost units are dollars per policy-year.
Appliance expected losses: \(1{,}200 \times \$38=\$45{,}600\).
Service-contract expected losses: \(900 \times \$24=\$21{,}600\).
Total expected losses: \(\$45{,}600+\$21{,}600=\$67{,}200\). Total earned premium is \(\$84{,}000+\$45{,}000=\$129{,}000\). Therefore, the implied combined expected loss ratio is \(\$67{,}200/\$129{,}000=0.52093\ldots\), which is 52.1% to three significant figures.
Check and mistake avoided. The total is consistent with the implied ratio: \(0.52093\ldots \times \$129{,}000=\$67{,}200\), subject to rounding. The two loss costs cannot be averaged and then applied to the combined exposure because they are different assumptions for different types of policy-year. Nor should either group’s loss cost be applied to the other group’s exposure.
Common Basis Errors
The arithmetic is usually simple; the harder part is ensuring that the inputs refer to the same risk volume and loss definition. The following errors can materially change an estimate even when the multiplication is performed correctly.
- Applying an expected loss ratio to the wrong premium base. A ratio defined for one premium measure should not be paired with another without establishing that the ratio is appropriate for it.
- Mixing exposure units. A cost per vehicle-year cannot be multiplied directly by policy count when policies have different vehicle counts or coverage durations.
- Using the wrong loss measure. An assumption for one coverage or loss definition does not automatically apply to a different coverage or definition.
- Averaging group assumptions without weights. Groups with different premium or exposure volumes contribute different amounts to the aggregate.
- Confusing a rate with a dollar amount. An expected loss ratio of 60% is \(0.60\) in multiplication, not 60; a loss cost of $80 per exposure unit is not itself the total expected loss.
Check Your Understanding
For each problem, identify the appropriate basis, show the key calculation, and check the units.
- A hypothetical coverage has $275,000 of matching earned premium and an expected loss ratio of 0.48. Find expected ultimate losses. (Answer: $132,000.)
- A hypothetical portfolio has 360 equipment-years and an expected loss cost of $72 per equipment-year. Find expected losses. (Answer: $25,920.)
- A program has 150 full-year units and 60 units insured for three months. Its expected loss cost is $96 per unit-year. Under a time-proportional convention, find total exposure and expected losses. (Answer: 165 unit-years and $15,840.)
- Two hypothetical groups have premiums of $100,000 and $300,000, with expected loss ratios of 0.40 and 0.60, respectively. Find total expected losses and the implied combined ratio; explain why the simple average is not the correct combined ratio. (Answer: $220,000 and 55.0%; the simple average of 50.0% ignores the different premium weights.)
- A hypothetical line has 500 matching exposure units, expected loss cost of $110 per unit, and matching premium of $100,000. Calculate expected losses and the implied expected loss ratio. State the assumption needed for that ratio to be meaningful. (Answer: $55,000 and 55.0%; the loss cost, exposure, premium, coverage, period, and loss definition must refer to the same risk basis.)