Why Expected Losses Must Match the Volume Being Estimated
An expected loss estimate applies to a specified amount of business. If the estimate is built for one earned-premium amount or one exposure count, it should not be carried unchanged to a period with a different volume. Scaling updates the estimate for that volume change while keeping the selected expected loss ratio or expected loss cost fixed.
The key is to use the same basis that produced the original expectation. A premium-based estimate scales with earned premium; an exposure-based estimate scales with the number of exposure units. This tutorial focuses on volume changes only. It does not adjust for a change in the composition of the exposures, which requires a separate analysis.
The Two Scaling Rules
For a premium-based estimate, the selected expected loss ratio is the expected loss per dollar of earned premium. If \(r\) is held fixed, multiplying earned premium by a factor multiplies expected ultimate losses by that same factor. For an exposure-based estimate, the selected expected loss cost is expected loss per exposure unit. If \(c\) is held fixed, multiplying the exposure count by a factor likewise multiplies expected ultimate losses by that factor.
The subscript 0 labels the original volume and 1 the volume being estimated. These formulas require the same coverage, loss and expense scope, period basis, and selected factor to apply at both volumes. For the exposure-based formula, the exposure unit must also be defined consistently. For example, vehicle-years cannot be interchanged with policies unless the relationship between those units is established.
Why Proportional Scaling Works
The premium-based result follows directly from the definition of a fixed expected loss ratio. At volume 0, expected losses are \(rP^e_0\). At volume 1, with the same ratio \(r\), they are \(rP^e_1\). If \(P^e_0>0\), substitute \(E[L^U_0]=rP^e_0\) into the ratio calculation:
The exposure-based result has the same justification: \(E[L^U_0]=cN_0\), so multiplying by \(N_1/N_0\) gives \(cN_1=E[L^U_1]\). No claim-size or claim-count distribution is needed for this scaling identity. It is an expected-value calculation using a fixed selected factor, not a claim that actual losses will increase by exactly the same proportion.
Determine whether the starting estimate was calculated using earned premium and an expected loss ratio, or exposure units and an expected loss cost.
Check that the scope and assumptions behind the expected loss ratio or loss cost still apply to the target volume.
Divide target earned premium by original earned premium, or target exposures by original exposures, using the matching basis.
Multiply the original expected loss by the volume factor. State the resulting expected ultimate loss and, when useful, the dollar change.
Worked Calculations
Worked Example: Scaling a Premium-Based Expectation
A hypothetical commercial property portfolio has an original earned premium of $2,400,000 and a selected expected loss ratio of 62%. A revised estimate is needed for $2,700,000 of earned premium. Assume the ratio applies unchanged to both premium amounts, and that the loss scope is identical. Find the revised expected ultimate loss and its increase from the original estimate.
Method and assumptions. This is premium-based scaling. The original and revised earned premium are on the same basis, and \(r=0.62\) is fixed. This is a deterministic expected-value calculation; no interest rate, mortality assumption, or claim distribution and parameters are used or required.
First calculate the original expectation:
Next calculate the premium volume factor:
Scale the original expected loss, or calculate directly using the fixed ratio:
and \(0.62(2{,}700{,}000)=1{,}674{,}000\), which checks the result. The increase is \(1{,}674{,}000-1{,}488{,}000=\$186{,}000\). The expected loss rises by 12.5%, the same percentage as earned premium.
Common mistake avoided: Scaling by the change in a different premium measure, such as written premium, would not follow from this calculation. The numerator must be the earned premium amounts that define the original and revised estimates.
Worked Example: Scaling an Exposure-Based Expectation
A hypothetical equipment insurer selected an expected loss cost of $184 per machine-year. An original estimate covers 8,500 machine-years; a revised estimate covers 9,200 machine-years. Calculate the revised expected ultimate loss and the change from the original estimate. Assume each exposure is the same defined unit and the selected loss cost and coverage scope are unchanged.
Method and assumptions. This is exposure-based scaling. The exposure count is in machine-years in both estimates. This expected-value calculation does not require an interest rate, mortality assumption, or claim distribution and parameters.
The original expectation is:
The exposure volume factor is:
Using the selected loss cost directly gives:
The difference is \(1{,}692{,}800-1{,}564{,}000=\$128{,}800\). As a check, \(1{,}564{,}000(92/85)=1{,}692{,}800\). Thus expected losses rise in proportion to machine-years, not by an independently chosen dollar adjustment.
Common mistake avoided: Multiplying by the change in exposure count, \(9{,}200-8{,}500=700\), and treating 700 as the total exposure volume would omit the original 8,500 units. The added expected loss is \(700(184)=\$128{,}800\); the revised total is the original estimate plus that increment.
Worked Example: Checking That Both Bases Agree
A hypothetical service-contract portfolio has 5,000 exposure units, $1,200,000 of earned premium, and a selected expected loss ratio of 65%. The premium per exposure unit is $240. A revised estimate has 6,250 exposure units, with the same $240 earned premium per unit. Calculate the revised expected loss using each basis, and check whether the results agree. Assume unchanged coverage and scope.
Method and assumptions. First derive the loss cost implied by the original premium and expected loss ratio. Then scale the premium-based and exposure-based estimates. Both bases should agree because premium per exposure and the selected ratio are held consistent. No interest rate, mortality assumption, or claim distribution and parameters are used.
The original premium-based expectation is:
The implied expected loss cost is \(780{,}000/5{,}000=\$156\) per exposure unit. The revised earned premium is \(6{,}250(240)=\$1{,}500{,}000\). The premium-based revised expectation is therefore:
The exposure-based revised expectation is:
The volume factor is \(6{,}250/5{,}000=1.25\); accordingly, \(780{,}000(1.25)=975{,}000\). Both methods agree. This check works because \(240(0.65)=156\), so the premium-based and exposure-based factors imply the same expected loss per exposure.
Common mistake avoided: A premium-based estimate and an exposure-based estimate should not be assumed to agree automatically. Their implied loss cost and premium per exposure must be compatible, as they are here.
Choosing the Correct Basis and Avoiding Traps
Volume scaling is simple only after the original basis is identified. The expected loss ratio and expected loss cost are not interchangeable numbers: the former is a proportion of premium, while the latter is a dollar amount per exposure unit. A change in one volume measure does not, by itself, establish the change in the other.
| Starting estimate | Fixed selected factor | Scale by | Check |
|---|---|---|---|
| Premium-based | Expected loss ratio \(r\) | Target earned premium / original earned premium | Both premium amounts use the same basis and scope |
| Exposure-based | Expected loss cost \(c\) | Target exposure units / original exposure units | Both counts use the same exposure unit and scope |
A common trap is to apply both scaling factors to one estimate. If an expected loss ratio is fixed and the earned premium has already been updated to reflect the target volume, calculate \(rP^e_1\); do not then multiply by an exposure factor as well. That would count the volume change twice. Conversely, if a loss cost is fixed, calculate \(cN_1\) and do not apply an additional premium factor.
Another trap is to assume that fixed expected loss ratio and fixed expected loss cost always produce the same answer. They do only when premium per exposure and the two selected factors are consistent. For example, if average earned premium per exposure changes while the loss cost is held constant, the expected loss ratio implied by that loss cost may change. An analyst must decide which selected basis is being held fixed rather than force the two methods to agree.
Worked Example: Detecting an Inconsistent Pair of Selections
A hypothetical policy group has 10,000 exposure units, $2,000,000 earned premium, and an expected loss ratio of 60%. An analyst also proposes a fixed expected loss cost of $125 per exposure. The target estimate keeps the exposure count at 10,000 but raises earned premium to $2,200,000. Compare the target expectations under the two stated fixed-factor assumptions and explain whether they agree. Assume the same loss scope.
Method and assumptions. Calculate each method separately: hold the loss ratio fixed for the premium-based estimate, then hold the loss cost fixed for the exposure-based estimate. No interest rate, mortality assumption, or claim distribution and parameters are involved.
The original expected loss from the 60% ratio is \(0.60(2{,}000{,}000)=\$1{,}200{,}000\), which implies a loss cost of \(1{,}200{,}000/10{,}000=\$120\), not $125. At the target premium, the fixed-ratio result is:
At the unchanged exposure count, the proposed fixed-$125 loss-cost result is:
The estimates differ by \(1{,}320{,}000-1{,}250{,}000=\$70{,}000\). They disagree because the original selections were already inconsistent: a 60% ratio applied to $200 of premium per exposure implies $120 of expected loss per exposure, whereas the proposed loss cost is $125. The answer is not to average the two estimates; it is to clarify which selection and basis governs.
Common mistake avoided: Using the target premium with the fixed ratio and also multiplying the result by an exposure factor of 1.00 does not resolve the inconsistency. Each method must be evaluated on its own stated basis.
Practical Checks
Before finalizing a scaled expectation, check that the volume factor is in the intended direction. If earned premium increases by 10% and the expected loss ratio stays fixed, expected ultimate loss must also increase by 10%. If exposures do not change and loss cost stays fixed, the exposure-based estimate must not change. These are useful reasonableness checks before rounding or reporting.
Also check whether the input is a baseline expected loss or an expected loss per unit. A baseline total is multiplied by a volume ratio. A selected loss cost is multiplied by the new exposure count directly. Confusing those forms can lead to multiplying by the exposure count twice. Keep the units visible in intermediate calculations: a loss ratio is dimensionless, earned premium is dollars, and loss cost is dollars per exposure unit.
Check Your Understanding
For each problem, use only the fixed factor and volume basis stated. Give dollar answers to the nearest cent if cents occur.
- A hypothetical liability book has $1,800,000 of earned premium and a selected expected loss ratio of 58%. Recalculate expected ultimate losses at $2,070,000 earned premium, keeping the ratio fixed. (Answer: $1,200,600, an increase of $156,600.)
- A hypothetical machinery portfolio has an expected loss cost of $92 per machine-year. Exposure volume rises from 12,400 to 13,150 machine-years. Find the revised expected loss and increase. (Answer: $1,209,800, an increase of $69,000.)
- An original premium-based expected loss is $2,160,000 at a selected expected loss ratio of 72%. Earned premium rises from its original amount to $3,450,000. Find the new expectation and the premium volume factor. (Answer: Original earned premium $3,000,000; factor 1.15; revised expected loss $2,484,000.)
- A hypothetical equipment plan has 4,800 exposure units and an expected loss cost of $135 per unit. Recalculate the expectation at 5,100 units and state the incremental expected loss. (Answer: $688,500, an increase of $40,500.)
- A portfolio has 10,000 exposure units, $2,000,000 earned premium, a fixed expected loss ratio of 60%, and a proposed fixed loss cost of $125 per unit. The target has the same exposure count and $2,200,000 earned premium. Calculate both target estimates and explain the discrepancy. (Answer: Fixed-ratio estimate $1,320,000; fixed-loss-cost estimate $1,250,000; they differ by $70,000 because the stated factors are inconsistent.)